Power load time-sharing prediction method, device, equipment and medium

Abstract

The invention discloses a power load time-sharing prediction method and device, equipment and a medium. The method comprises the following steps: respectively carrying out time period segmentation on data of a to-be-predicted day and each historical similar day; determining each reference similar day of the current time period according to the meteorological similarity between the to-be-predicted meteorological characteristics of the current time period and each historical meteorological characteristic of the current time period; decomposing the power load sequence of the current time period in each reference similar day to obtain each first power load sub-sequence of the current time period; on the basis of a power load prediction model, according to the reference meteorological characteristics of the current time period in the reference similar day and the first power load subsequences of the current time period, determining the second power load subsequences of the current time period; determining a fusion power load sequence of the current time period according to the second power load subsequences of the current time period; and obtaining a power load prediction sequence of the to-be-predicted day according to the fused power load sequence of each current time period. According to the embodiment of the invention, the power load prediction precision can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power load forecasting, and in particular to a method, device, equipment and medium for time-sharing forecasting of power load. Background Art

[0002] Power system load forecasting is of great significance for the reasonable arrangement of dispatching plans and improving the economy of system operation. The power market is gradually developing and improving, and higher requirements are placed on the timeliness and accuracy of load forecasting. Therefore, short-term load forecasting has become an important research topic in the power field. Over the years, various load forecasting theories and methods have been proposed in China, including traditional forecasting methods such as time series method, exponential smoothing method, and gray forecasting method. Traditional forecasting methods generally have high requirements for historical data, and it is difficult to consider various factors that affect load changes, resulting in a relatively simple model structure and low prediction accuracy. Summary of the invention

[0003] The present invention provides a method, device, equipment and medium for time-sharing prediction of electric load, so as to solve the problem that the prediction accuracy of the existing electric load prediction method is low.

[0004] According to one aspect of the present invention, a method for time-sharing prediction of electric load is provided, comprising:

[0005] Determine each historical similar day of the day to be predicted, divide the data of the day to be predicted and each historical similar day according to time periods, and obtain the meteorological characteristics of the day to be predicted and each historical meteorological characteristics of each historical similar day in the current time period;

[0006] Determine the meteorological similarity between the meteorological characteristics to be predicted in the current period and the historical meteorological characteristics in the current period, and determine the reference similar days in the current period from the historical similar days according to the meteorological similarity;

[0007] The variational mode decomposition algorithm is used to decompose the power load sequence of the current period in each reference similar day to obtain each first power load subsequence of the current period;

[0008] Based on the power load forecasting model, determining each second power load subsequence of the current period according to the reference meteorological characteristics of the current period in the reference similar day and each first power load subsequence of the current period;

[0009] Fusing each second power load subsequence in the current period to determine a fused power load sequence in the current period;

[0010] The fused power load sequences of each current time period are spliced ​​to obtain the power load forecast sequence of the day to be predicted.

[0011] According to another aspect of the present invention, there is provided a device for time-sharing prediction of electric load, comprising:

[0012] The time segmentation module is used to determine the historical similar days of the day to be predicted, and to segment the data of the day to be predicted and the historical similar days according to the time periods, so as to obtain the meteorological characteristics of the day to be predicted and the historical meteorological characteristics of the historical similar days in the current time period;

[0013] A reference similar day determination module is used to determine the meteorological similarity between the meteorological characteristics to be predicted in the current period and the historical meteorological characteristics in the current period, and determine the reference similar days in the current period from the historical similar days according to the meteorological similarity;

[0014] The power load sequence decomposition module is used to decompose the power load sequence of the current period in each reference similar day by using a variational mode decomposition algorithm to obtain each first power load subsequence of the current period;

[0015] A subsequence prediction module, for determining each second power load subsequence of the current period based on the power load prediction model, according to the reference meteorological characteristics of the current period in the reference similar day and each first power load subsequence of the current period;

[0016] A subsequence fusion module, used to fuse each second power load subsequence of the current period to determine a fused power load sequence of the current period;

[0017] The time period sequence splicing module is used to splice the fused power load sequences of each current time period to obtain the power load forecast sequence of the day to be predicted.

[0018] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method for time-sharing prediction of electric load according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0020] At least one processor; and a memory in communication with the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for time-sharing prediction of electric load as described in any embodiment of the present invention.

[0021] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the power load time-sharing prediction method described in any embodiment of the present invention when executed.

[0022] The embodiment of the present invention divides the whole day into different time periods, predicts the power load of each time period according to the meteorological characteristics of each time period to obtain the fused power load sequence of each time period, and then splices the fused power load sequence of each time period in chronological order to obtain the power load prediction sequence of the day to be predicted, and fully refers to the local correlation between the day to be predicted and similar historical days in each time period. The meteorological data of different time periods have a significant impact on the power load. By dividing the whole day into multiple time periods and predicting the meteorological characteristics of each time period, these influencing factors can be captured more accurately, thereby improving the accuracy of power load prediction.

[0023] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 This is a first flow chart of a method for time-sharing prediction of electric load provided by an embodiment of the present invention;

[0026] Figure 2 is a second flow chart of a method for time-sharing prediction of electric load provided by an embodiment of the present invention;

[0027] Figure 3 It is a structural schematic diagram of a power load time-sharing prediction device provided by an embodiment of the present invention;

[0028] Figure 4 It is a schematic diagram of the structure of an electronic device implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] Figure 1 This is the first flow chart of a method for time-sharing prediction of electric load provided by an embodiment of the present invention. This embodiment is applicable to the case of predicting electric load by time period. The method can be executed by a device for time-sharing prediction of electric load. The device can be implemented in the form of hardware and / or software. The device can be configured in an electronic device with corresponding data processing capabilities. Figure 1 As shown, the method includes:

[0032] S110, determining the historical similar days of the day to be predicted, dividing the data of the day to be predicted and the historical similar days according to time periods, and obtaining the meteorological characteristics of the day to be predicted and the historical meteorological characteristics of the historical similar days in the current time period.

[0033] Determine the historical similar days of the day to be predicted, and divide the day to be predicted and the historical similar days into multiple time periods. For example, the whole day can be divided into six time periods: time period 1 (0-4 o'clock), time period 2 (4-8 o'clock), time period 3 (8-12 o'clock), time period 4 (12-16 o'clock), time period 5 (16-20 o'clock) and time period 6 (20-24 o'clock). For each time period, the meteorological features to be predicted in the current time period are constructed for the day to be predicted based on the meteorological data of the day to be predicted in the current time period, and the historical meteorological features of the current time period are constructed for each historical similar day based on the meteorological data of each historical similar day in the current time period.

[0034] Optionally, for the day to be predicted, the average temperature, maximum temperature, minimum temperature, average humidity, maximum humidity, average rainfall, rainfall duration, average snowfall and snowfall duration of the current period are selected to construct the feature vector to be predicted for the current period, and the feature vector to be predicted is normalized to obtain the meteorological characteristics to be predicted for the day to be predicted in the current period. For historically similar days, the average temperature, maximum temperature, minimum temperature, average humidity, maximum humidity, average rainfall, rainfall duration, average snowfall and snowfall duration of the current period are selected to construct the period historical feature vector of the current period, and the period historical feature vector is normalized to obtain the historical meteorological characteristics of historically similar days in the current period.

[0035] S120, determining the meteorological similarity between the meteorological features to be predicted in the current period and the historical meteorological features in the current period, and determining the reference similar days in the current period from the historical similar days according to the meteorological similarity.

[0036] The meteorological similarities between the meteorological features to be predicted in the current period and the various historical meteorological features in the current period are determined respectively, and a preset number of historical similar days with the highest meteorological similarities are selected as reference similar days for the current period.

[0037] S130. Decompose the power load sequence of the current time period in each reference similar day using a variational mode decomposition algorithm to obtain each first power load subsequence of the current time period.

[0038] Modal decomposition is to decompose a set of data into a set of components and residual data. The variational modal decomposition algorithm is used to decompose the power load sequence of the current period in each reference similar day to obtain each first power load subsequence of the current period. Optionally, the adaptive particle swarm optimization algorithm is used to optimize the number of decomposition modes and the quadratic penalty factor in the variational modal decomposition algorithm parameters, and the optimized variational modal decomposition algorithm is used to decompose the power load sequence of the current period in each reference similar day to obtain each first power load subsequence of the current period. The optimized variational modal decomposition algorithm can more accurately capture the different frequency components in the power load sequence, avoid problems such as modal aliasing and over-decomposition, and the decomposed first power load subsequences are clearer and more accurate, providing a good foundation for subsequent analysis and prediction.

[0039] S140. Based on the power load prediction model, determine each second power load subsequence of the current time period according to the reference meteorological characteristics of the current time period in the reference similar day and each first power load subsequence of the current time period.

[0040] The meteorological similarity between each historical meteorological feature of the current period of each reference similar day and the meteorological feature to be predicted in the current period is determined respectively, and the historical meteorological feature with the highest meteorological similarity is used as the reference meteorological feature of the current period. The reference meteorological feature of the current period and each first power load subsequence of the current period are respectively used as the input of each power load prediction model, and each second power load subsequence of the current period is predicted according to each power load prediction model. One first power load subsequence corresponds to one power load prediction model to obtain one second power load subsequence.

[0041] Through variational mode decomposition, complex power load sequences are converted into multiple relatively simple power load subsequences, making the prediction process more accurate. Separate prediction of each power load subsequence can fully consider the unique properties of each component, thereby avoiding the model mismatch problem that may occur in traditional prediction methods. Targeted prediction methods help improve the accuracy of predictions. By predicting each power load subsequence, local features in the power load sequence, such as spikes and mutations, can be captured more carefully. After decomposing the power load sequence into multiple power load subsequences, a relatively simple prediction model can be established for each power load subsequence, which can reduce the complexity of the model and improve computational efficiency. Since the prediction process of each power load subsequence is independent, parallel processing can be used to further improve the calculation speed.

[0042] S150, fusing each second power load subsequence of the current period to determine a fused power load sequence of the current period; splicing the fused power load sequences of each current period to obtain a power load forecast sequence for the day to be forecasted.

[0043] The second power load subsequences of the current period are superimposed to determine the fused power load sequence of the current period. The fused power load sequence of each current period is obtained, and the fused power load sequences of each current period are spliced ​​in chronological order to obtain the power load forecast sequence of the day to be forecasted.

[0044] The embodiment of the present invention divides the whole day into different time periods, predicts the power load of each time period according to the meteorological characteristics of each time period to obtain the fused power load sequence of each time period, and then splices the fused power load sequence of each time period in chronological order to obtain the power load prediction sequence of the day to be predicted, and fully refers to the local correlation between the day to be predicted and similar historical days in each time period. The meteorological data of different time periods have a significant impact on the power load. By dividing the whole day into multiple time periods and predicting the meteorological characteristics of each time period, these influencing factors can be captured more accurately, thereby improving the accuracy of power load prediction.

[0045] In an optional embodiment, the method also includes filling in the missing values ​​in the power load data of similar historical days. Specifically, determining the moment at which the missing value is located in the current similar historical day as the current moment; taking the average value of the power load value at the current moment in other similar historical days as the first reference value; taking the average value of the power load values ​​corresponding to other moments in the current similar historical day as the second reference value; taking the average value of the power load values ​​corresponding to other moments in other similar historical days as the third reference value; and determining the missing value in the power load data of the current similar historical day based on the first reference value, the second reference value and the third reference value.

[0046] The historical similar day with missing values ​​in the power load data is taken as the current historical similar day, and the moment where the missing value is located in the current historical similar day is determined as the current moment. The average value of the power load value at the current moment in other historical similar days is taken as the first reference value. The average value of the power load value corresponding to other moments in the current historical similar day is taken as the second reference value. The average value of the power load value corresponding to other moments in other historical similar days is taken as the third reference value. The sum of the first reference value and the second reference value minus the third reference value is used as the filling value, and the power load data at the current moment in the current historical similar day is filled according to the filling value. By calculating the average power load value at the current moment in other historical similar days (first reference value) and the average power load value at other moments in the current historical similar day (second reference value), and the average power load value at other moments in other historical similar days (third reference value), the power load data of multiple historical similar days and multiple moments are comprehensively considered, and the information in the historical data can be more comprehensively utilized to improve the accuracy and reliability of the filling value. The time change and spatial distribution characteristics of the power load data are fully considered, which helps to maintain the coherence and consistency of the filled data in time and space, and improve the data quality.

[0047] In an optional embodiment, the method also includes: for the power load data at each moment of a similar historical day, judging whether there is a load mutation in the power load value at the current moment based on the judgment coefficient and the power load value at the previous moment; if there is a load mutation in the power load value at the current moment and the current moment is an intermediate moment, determining the corrected power load value at the current moment based on the power load value at the previous moment and the power load value at the next moment; if there is a load mutation in the power load value at the current moment and the current moment is the last moment, determining the corrected power load value at the current moment based on the power load values ​​at the first two moments adjacent to the current moment.

[0048] Specifically, as shown in the following formula (1), if the power load value at the current moment and the power load value at the previous moment on a similar historical day meet any of the conditions in formula (1), it is determined that there is a sudden load change in the power load value at the current moment.

[0049]

[0050] Among them, L(i) is the power load value at time i; α is the judgment coefficient; n is the total number of moments divided by the power load data of similar historical days.

[0051] If there is a sudden load change in the power load value at the current moment and the current moment is an intermediate moment, the sudden load change is an intermediate load change, and the corrected power load value at the current moment is determined according to the power load value at the moment before the current moment and the power load value at the moment after the current moment, as shown in the following formula (2). If there is a sudden load change in the power load value at the current moment and the current moment is the last moment, the sudden load change is a tail load change, and the corrected power load value at the current moment is determined according to the power load values ​​at the first two moments adjacent to the current moment, as shown in the following formula (3).

[0052] L(i)=0.5×[L(i-1)+L(i+1)],i=2,3,...,n-1 (2)

[0053]

[0054] Where n is the total number of moments at which the power load data of similar historical days are divided.

[0055] By correcting the load mutations in the power load data of similar historical days and adopting different correction methods according to the load conditions at different time points, the accuracy of the power load data is improved, thereby improving the accuracy of the power load time-sharing forecast.

[0056] Figure 2 This is a second flow chart of a method for time-sharing prediction of power load provided by an embodiment of the present invention. This embodiment is optimized and improved on the basis of the above embodiment. Figure 2 As shown, the method includes:

[0057] S210: Determine the weather type of the day to be predicted according to the weather forecast, and select a historical day with the same weather type as the day to be predicted as a first similar day.

[0058] Specifically, seasons include spring, summer, autumn and winter; weather includes sunny, cloudy, rainy and snowy days. Each season and each weather is combined to determine sixteen weather types. The weather type of the day to be predicted is determined according to the weather forecast. From each historical day, a historical day with the same weather type as the day to be predicted is selected as the first similar day.

[0059] S220 , determining historical similar days of the day to be predicted from the first similar days according to the similarity between the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of each first similar day.

[0060] The whole-day feature vector of the day to be predicted is constructed based on the average temperature, maximum temperature, minimum temperature, average humidity, maximum humidity, average rainfall, rainfall duration, average snowfall and snowfall duration of the whole day of the day to be predicted, and the whole-day feature vector is normalized to obtain the whole-day meteorological characteristics of the day to be predicted. The whole-day historical feature vector of the first similar day is constructed based on the average temperature, maximum temperature, minimum temperature, average humidity, maximum humidity, average rainfall, rainfall duration, average snowfall and snowfall duration of each first similar day, and the whole-day historical feature vector is normalized to obtain the whole-day meteorological characteristics of the first similar day.

[0061] The similarity between the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of each first similar day is determined, and a set number of first similar days with the highest similarity are selected from each first similar day as historical similar days according to the similarity.

[0062] In an optional implementation manner, determining the historical similar days of the day to be predicted from the first similar days according to the similarity between the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of each first similar day, comprises:

[0063] The maximum reference factor and the minimum reference factor of the current first similar day are determined according to the all-day meteorological characteristics of the day to be predicted, the all-day meteorological characteristics of the current first similar day, and the all-day meteorological characteristics of each first similar day; wherein the all-day meteorological characteristics include at least one characteristic component.

[0064] Optionally, the overall meteorological characteristics include nine characteristic components: average temperature, maximum temperature, minimum temperature, average humidity, maximum humidity, average rainfall, rainfall duration, average snowfall and snowfall duration.

[0065] The first similarity between the day to be predicted and the current first similar day is determined according to the maximum reference factor and the minimum reference factor of the current first similar day, the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of the current first similar day.

[0066] Specifically, the correlation coefficient of any characteristic component between the day to be predicted and the current first similar day is determined. The specific process is as follows: Formula (4).

[0067]

[0068] Among them, A i is the minimum reference factor of the ith first similar day; B iis the maximum reference factor of the first similar day of the ith time period; ρ is a coefficient ranging from 0 to 1, with a default value of 0.5; X p (k) is the kth characteristic component of the day to be predicted; X i (k) is the kth characteristic component of the i-th first phase day; ψ i (k) is the correlation coefficient of the kth characteristic component between the day to be predicted and the i-th first similar day.

[0069] The correlation coefficients of the characteristic components between the day to be predicted and the current first similar day are accumulated to determine the first similarity between the day to be predicted and the current first similar day. The specific process is shown in the following formula (5).

[0070]

[0071] Where N represents the number of characteristic components; ξ 1-i Represents the first similarity between the i-th first similar day and the day to be predicted.

[0072] The second similarity between the day to be predicted and the current first similar day is determined according to the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of the current first similar day. The specific calculation process is shown in the following formula (6).

[0073]

[0074] The weighted similarity between the to-be-predicted day and the current first similar day is determined according to the first similarity and the second similarity. Specifically, as shown in the following formula (7).

[0075] ξ i =σξ 1-i +(1-σ)ξ 2-i (7)

[0076] Among them, σ is a weight coefficient with a value between 0 and 1, which can be selected according to the specific weather type of the day to be predicted. When the temperature, humidity, rainfall, etc. fluctuate violently, that is, the weather type changes significantly, σ can be taken to be close to 0. When the temperature, humidity, rainfall, etc. fluctuate slightly, that is, the weather type basically does not change, σ can be taken to be close to 1.

[0077] According to the weighted similarity, a set number of first similar days with the highest weighted similarity are selected from the first similar days as historical similar days.

[0078] By first determining the first similarity and the second similarity, and then calculating the weighted similarity using the first similarity and the second similarity, it is possible to select a first similar day whose meteorological factors are more similar to those of the day to be predicted from the first similar days, thereby improving the accuracy of the power load forecast results for subsequent days to be predicted.

[0079] In an optional embodiment, the method of determining the maximum reference factor and the minimum reference factor of the current first similar day based on the all-day meteorological characteristics of the day to be predicted, the all-day meteorological characteristics of the current first similar day, and the all-day meteorological characteristics of each first similar day includes: determining the difference between each characteristic component of the day to be predicted and the corresponding characteristic component of the current first similar day as the component difference of the current characteristic component; taking the characteristic component corresponding to the minimum value in the component differences of the characteristic components of the day to be predicted as the first characteristic component; taking the characteristic component corresponding to the maximum value in the component differences of the characteristic components of the day to be predicted as the second characteristic component; determining the difference between the day to be predicted and each first similar day in the first characteristic component, and taking the minimum value in the difference as the minimum reference factor of the current first similar day; determining the difference between the day to be predicted and each first similar day in the second characteristic component, and taking the maximum value in the difference as the maximum reference factor of the current first similar day.

[0080] Specifically, it is shown in the following formulas (8) and (9).

[0081]

[0082] Where T is the first characteristic component; F is the second characteristic component; X p (t) is the t-th characteristic component of the day to be predicted; X i (t) is the t-th characteristic component of the i-th first phase day; X p (T) is the first characteristic component of the day to be predicted; X r (T) is the first characteristic component of the first phase day of the rth phase; X p (F) is the second characteristic component of the day to be predicted; X r (T) is the second characteristic component of the rth first phase day; A i is the minimum reference factor of the ith first similar day; B i is the maximum reference factor of the i-th first similar day.

[0083] S230, dividing the data of the day to be predicted and the data of each similar historical day according to time periods, and obtaining the meteorological characteristics of the day to be predicted and each historical meteorological characteristics of each similar historical day in the current time period.

[0084] S240: Determine the meteorological similarity between the meteorological features to be predicted in the current period and the historical meteorological features in the current period, and determine the reference similar days in the current period from the historical similar days according to the meteorological similarity.

[0085] The meteorological similarity between the meteorological features to be predicted in the current period and the historical meteorological features in the current period can be determined by referring to the similarity calculation method between the meteorological features of the whole day.

[0086] S250, using a variational mode decomposition algorithm to decompose the power load sequence of the current time period in each reference similar day to obtain each first power load subsequence of the current time period.

[0087] S260. Based on the power load prediction model, determine each second power load subsequence of the current time period according to the reference meteorological characteristics of the current time period in the reference similar day and each first power load subsequence of the current time period.

[0088] Optionally, the power load forecasting model may select an LSTM (Long Short-Term Memory) model or a GRU (Gated Recurrent Unit).

[0089] In an optional implementation, the power load forecasting model is trained based on a sparrow search algorithm.

[0090] Specifically, when the power load forecasting model is an LSTM model, the power load forecasting model is trained through the following process:

[0091] 1) Initialize the number of hidden layer neurons in the power load forecasting model, the number of model training iterations, and the optimization range of the learning rate σ;

[0092] 2) Initialize the sparrow size, the ratio of discoverers, joiners and guards in the sparrow search algorithm, and the number of algorithm iterations;

[0093] 3) Calculate the fitness value of the sparrow search algorithm to obtain the optimal fitness value, the worst fitness value and their corresponding positions of the group;

[0094] 4) Update the positions of discoverers, joiners, and alerters in the sparrow population;

[0095] 5) Update the optimal position. If the iteration completion condition is met, proceed to the next step. If not, return to step 3) and continue searching until the iteration completion condition is met.

[0096] 6) The number of hidden layer neurons corresponding to the optimal position when the iteration completion condition is met, the number of model training iterations, and the learning rate σ are used as the number of hidden layer neurons, the number of model training iterations, and the learning rate of the power load forecasting model, respectively, to obtain the power load forecasting model.

[0097] Using the sparrow search algorithm to optimize the parameters of the power load forecasting model can not only improve the prediction efficiency of the model, but also improve the prediction accuracy of the model.

[0098] S270, fusing each second power load subsequence of the current period to determine a fused power load sequence of the current period; and splicing the fused power load sequences of each current period to obtain a power load forecast sequence for the day to be forecasted.

[0099] The embodiment of the present invention determines the weather type of the day to be predicted according to the weather forecast, selects the historical day with the same weather type as the day to be predicted as the first similar day; determines the historical similar day of the day to be predicted from the first similar day according to the similarity between the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of each first similar day. It is possible to select a historical similar day with a strong similarity to the meteorological factors of the day to be predicted from the historical days, thereby improving the accuracy of the power load prediction results of the subsequent day to be predicted. By dividing the whole day into different time periods, the power load of each time period is predicted according to the meteorological characteristics of each time period to obtain the fused power load sequence of each time period, and then the fused power load sequence of each time period is spliced ​​in time order to obtain the power load prediction sequence of the day to be predicted, fully referring to the local correlation between the day to be predicted and the historical similar day in each time period, and referring to the significant impact of meteorological data in different time periods on the power load, by dividing the whole day into multiple time periods, and predicting the meteorological characteristics of each time period respectively, these influencing factors can be captured more accurately, thereby improving the accuracy of power load prediction.

[0100] Figure 3 It is a structural schematic diagram of a power load time-sharing prediction device provided by an embodiment of the present invention.

[0101] like Figure 3 As shown, the device comprises:

[0102] The time segmentation module 310 is used to determine the historical similar days of the day to be predicted, and to segment the data of the day to be predicted and the historical similar days according to the time periods, so as to obtain the meteorological characteristics to be predicted of the day to be predicted and the historical meteorological characteristics of the historical similar days in the current time period;

[0103] A reference similar day determination module 320 is used to determine the meteorological similarity between the meteorological features to be predicted in the current period and the historical meteorological features in the current period, and determine the reference similar days in the current period from the historical similar days according to the meteorological similarity;

[0104] The power load sequence decomposition module 330 is used to decompose the power load sequence of the current period in each reference similar day by using a variational mode decomposition algorithm to obtain each first power load subsequence of the current period;

[0105] The subsequence prediction module 340 is used to determine each second power load subsequence of the current period based on the power load prediction model according to the reference meteorological characteristics of the current period in the reference similar day and each first power load subsequence of the current period;

[0106] The subsequence fusion module 350 is used to fuse the second power load subsequences of the current period to determine the fused power load sequence of the current period;

[0107] The time period sequence splicing module 360 ​​is used to splice the fused power load sequences of each current time period to obtain the power load forecast sequence of the day to be forecasted.

[0108] The power load time-sharing prediction device provided in the embodiment of the present invention can execute the power load time-sharing prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0109] Optionally, the time segmentation module includes:

[0110] A first similar day determining unit, configured to determine the weather type of the day to be predicted according to the weather forecast, and select a historical day with the same weather type as the day to be predicted as the first similar day;

[0111] The historical similar day determination unit is used to determine the historical similar days of the day to be predicted from the first similar days according to the similarity between the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of each first similar day.

[0112] Optionally, the historical similar day determination unit includes:

[0113] A reference factor determination subunit, configured to determine the maximum reference factor and the minimum reference factor of the current first similar day according to the whole-day meteorological characteristics of the day to be predicted, the whole-day meteorological characteristics of the current first similar day, and the whole-day meteorological characteristics of each first similar day; wherein the whole-day meteorological characteristics include at least one characteristic component;

[0114] A first similarity determination subunit is used to determine a first similarity between the to-be-predicted day and the current first similar day according to the maximum reference factor and the minimum reference factor of the current first similar day, the whole-day meteorological characteristics of the to-be-predicted day, and the whole-day meteorological characteristics of the current first similar day;

[0115] A second similarity determination subunit is used to determine a second similarity between the day to be predicted and the current first similar day according to the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of the current first similar day;

[0116] A weighted similarity determination subunit, configured to determine a weighted similarity between the to-be-predicted day and a current first similar day according to the first similarity and the second similarity;

[0117] The historical similar day determination subunit is used to select a set number of first similar days with the highest weighted similarity from the first similar days as the historical similar days according to the weighted similarity.

[0118] Optionally, the reference factor determination subunit is specifically used for:

[0119] Determine the difference between each characteristic component of the day to be predicted and the characteristic component corresponding to the current first similar day as the component difference of the current characteristic component;

[0120] The characteristic component corresponding to the minimum value of the component differences of the characteristic components of the day to be predicted is taken as the first characteristic component;

[0121] The characteristic component corresponding to the maximum value of the component differences of the characteristic components of the day to be predicted is used as the second characteristic component;

[0122] Determine the difference between the to-be-predicted day and each first similar day in the first characteristic component, and use the minimum value of the difference as the minimum reference factor of the current first similar day;

[0123] The difference between the day to be predicted and each first similar day in the second characteristic component is determined, and the maximum value of the difference is used as the maximum reference factor of the current first similar day.

[0124] Optionally, the device also includes a missing value filling module, which is used to determine the moment where the missing value is located in the current historical similar day as the current moment; take the average value of the power load value at the current moment in other historical similar days as the first reference value; take the average value of the power load values ​​corresponding to other moments in the current historical similar day as the second reference value; take the average value of the power load values ​​corresponding to other moments in other historical similar days as the third reference value; and determine the missing value in the power load data of the current historical similar day based on the first reference value, the second reference value and the third reference value.

[0125] Optionally, the device also includes: a load mutation correction module, which is used to judge whether there is a load mutation in the power load value at the current moment based on the judgment coefficient and the power load value at the previous moment for the power load data at each moment on similar historical days; if there is a load mutation in the power load value at the current moment and the current moment is an intermediate moment, the corrected power load value at the current moment is determined based on the power load value at the previous moment and the power load value at the next moment; if there is a load mutation in the power load value at the current moment and the current moment is the last moment, the corrected power load value at the current moment is determined based on the power load values ​​at the first two moments adjacent to the current moment.

[0126] Optionally, the power load forecasting model is trained based on a sparrow search algorithm.

[0127] The power load time-sharing prediction device further described can also execute the power load time-sharing prediction method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0128] Figure 4 A schematic diagram of the structure of an electronic device 40 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0129] like Figure 4 As shown, the electronic device 40 includes at least one processor 41, and a memory connected to the at least one processor 41, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 to the random access memory (RAM) 43. In the RAM 43, various programs and data required for the operation of the electronic device 40 can also be stored. The processor 41, the ROM 42, and the RAM 43 are connected to each other through a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0130] A number of components in the electronic device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the electronic device 40 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0131] The processor 41 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 41 executes the various methods and processes described above, such as a power load time-sharing prediction method.

[0132] In some embodiments, the power load time-sharing prediction method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 40 via the ROM 42 and / or the communication unit 49. When the computer program is loaded into the RAM 43 and executed by the processor 41, one or more steps of the power load time-sharing prediction method described above may be performed. Alternatively, in other embodiments, the processor 41 may be configured to perform the power load time-sharing prediction method in any other appropriate manner (e.g., by means of firmware).

[0133] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0134] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0135] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0136] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0137] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0138] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0139] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0140] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for time-sharing prediction of power load, characterized in that: The method comprises: Determine each historical similar day of the day to be predicted, divide the data of the day to be predicted and each historical similar day according to time periods, and obtain the meteorological characteristics of the day to be predicted and each historical meteorological characteristics of each historical similar day in the current time period; Determine the meteorological similarity between the meteorological characteristics to be predicted in the current period and the historical meteorological characteristics in the current period, and determine the reference similar days in the current period from the historical similar days according to the meteorological similarity; The variational mode decomposition algorithm is used to decompose the power load sequence of the current period in each reference similar day to obtain each first power load subsequence of the current period; Based on the power load forecasting model, determining each second power load subsequence of the current period according to the reference meteorological characteristics of the current period in the reference similar day and each first power load subsequence of the current period; Fusing each second power load subsequence in the current period to determine a fused power load sequence in the current period; The fused power load sequences of each current time period are spliced ​​to obtain the power load forecast sequence of the day to be predicted.

2. The method according to claim 1, characterized in that The historical similar days for determining the day to be predicted include: Determine the weather type of the day to be predicted according to the weather forecast, and select a historical day with the same weather type as the day to be predicted as a first similar day; According to the similarity between the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of each first similar day, the historical similar days of the day to be predicted are determined from the first similar days.

3. The method according to claim 2, characterized in that The determining of historical similar days to the day to be predicted from the first similar days according to the similarity between the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of each first similar day, comprises: Determine the maximum reference factor and the minimum reference factor of the current first similar day according to the all-day meteorological characteristics of the day to be predicted, the all-day meteorological characteristics of the current first similar day, and the all-day meteorological characteristics of each first similar day; wherein the all-day meteorological characteristics include at least one characteristic component; Determine a first similarity between the day to be predicted and the current first similar day according to the maximum reference factor and the minimum reference factor of the current first similar day, the whole-day meteorological characteristics of the day to be predicted, and the whole-day meteorological characteristics of the current first similar day; Determining a second similarity between the day to be predicted and the current first similar day according to the whole-day meteorological characteristics of the day to be predicted and the whole-day meteorological characteristics of the current first similar day; Determining a weighted similarity between the to-be-predicted day and the current first similar day according to the first similarity and the second similarity; According to the weighted similarity, a set number of first similar days with the highest weighted similarity are selected from the first similar days as historical similar days.

4. The method according to claim 3, characterized in that The step of determining the maximum reference factor and the minimum reference factor of the current first similar day according to the all-day meteorological characteristics of the day to be predicted, the all-day meteorological characteristics of the current first similar day, and the all-day meteorological characteristics of each first similar day includes: Determine the difference between each characteristic component of the day to be predicted and the characteristic component corresponding to the current first similar day as the component difference of the current characteristic component; The characteristic component corresponding to the minimum value of the component differences of the characteristic components of the day to be predicted is taken as the first characteristic component; The characteristic component corresponding to the maximum value of the component differences of the characteristic components of the day to be predicted is used as the second characteristic component; Determine the difference between the to-be-predicted day and each first similar day in the first characteristic component, and use the minimum value of the difference as the minimum reference factor of the current first similar day; The difference between the day to be predicted and each first similar day in the second characteristic component is determined, and the maximum value of the difference is used as the maximum reference factor of the current first similar day.

5. The method according to claim 1, characterized in that The method further comprises: Determine the moment where the missing value is located in the current historical similar day as the current moment; Taking the average value of the power load value at the current time on other similar historical days as the first reference value; The average value of the power load values ​​corresponding to other times in the current historical similar day is used as the second reference value; The average value of the power load values ​​corresponding to other times on other similar historical days is taken as the third reference value; The missing value in the power load data of the current historical similar day is determined according to the first reference value, the second reference value and the third reference value.

6. The method according to claim 1, characterized in that The method further comprises: For the power load data at each time of similar historical days, determine whether there is a sudden change in the power load value at the current time based on the judgment coefficient and the power load value at the previous time; If there is a sudden change in the power load value at the current moment and the current moment is an intermediate moment, the corrected power load value at the current moment is determined according to the power load value at the moment before the current moment and the power load value at the moment after the current moment; If there is a sudden load change in the power load value at the current moment and the current moment is the last moment, the corrected power load value at the current moment is determined based on the power load values ​​at the previous two moments adjacent to the current moment.

7. The method according to claim 1, characterized in that The power load forecasting model is obtained by training based on a sparrow search algorithm.

8. A time-sharing prediction device for power load, characterized in that: The device comprises: The time segmentation module is used to determine the historical similar days of the day to be predicted, and to segment the data of the day to be predicted and the historical similar days according to the time periods, so as to obtain the meteorological characteristics of the day to be predicted and the historical meteorological characteristics of the historical similar days in the current time period; A reference similar day determination module is used to determine the meteorological similarity between the meteorological characteristics to be predicted in the current period and the historical meteorological characteristics in the current period, and determine the reference similar days in the current period from the historical similar days according to the meteorological similarity; The power load sequence decomposition module is used to decompose the power load sequence of the current period in each reference similar day by using a variational mode decomposition algorithm to obtain each first power load subsequence of the current period; A subsequence prediction module, for determining each second power load subsequence of the current period based on the power load prediction model, according to the reference meteorological characteristics of the current period in the reference similar day and each first power load subsequence of the current period; A subsequence fusion module, used to fuse each second power load subsequence of the current period to determine a fused power load sequence of the current period; The time period sequence splicing module is used to splice the fused power load sequences of each current time period to obtain the power load forecast sequence of the day to be predicted.

9. An electronic device, characterized in that: The electronic device comprises: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for time-sharing prediction of electric load according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the power load time-sharing prediction method according to any one of claims 1 to 7 when executed.

Images