A load prediction method and related device

Abstract

This application provides a load forecasting method and related equipment for predicting short-term load. The method includes: acquiring target sample data of users' historical electricity consumption, the target sample data including multiple target daily electricity consumption data and target time-of-use electricity consumption data; inputting the target daily electricity consumption data into a pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily average electricity consumption data; inputting the multiple sets of intermediate typical daily average electricity consumption data into a pre-trained linear regression model to obtain target typical daily average electricity consumption data; clustering the target time-of-use electricity consumption data and the target daily electricity consumption data corresponding to the target time-of-use electricity consumption data using a clustering regression model to obtain type-day time-of-use weight coefficient data; and predicting short-term load using the target typical daily average electricity consumption data, typical daily coefficient data, and type-day time-of-use weight coefficient data.

Description

Technical Field

[0001] This application relates to the field of day-ahead load forecasting in the electricity spot market, and more particularly to a load forecasting method and related equipment. Background Technology

[0002] Load forecasting methods in power system applications can be categorized based on the forecast duration: long-term, medium-term, and short-term load forecasting methods. Among classic short-term load forecasting methods, the most widely used include regression analysis, time series analysis, grey forecasting, and Kalman filtering. For example, the paper "Short-Term Load Forecasting Model Based on Online Sequence Limit Support Vector Regression" utilizes online regression analysis to build a short-term load forecasting model and validates its effectiveness using data from New England Conservative Schools. Another example is "Short-Term Load Forecasting of Power Systems Based on Time Series Analysis and Kalman Filtering," which combines time series analysis and Kalman filtering to improve forecast accuracy and overcome the forecast delay inherent in time series analysis.

[0003] As the electricity market gradually transitions into a long-term spot market, short-term load forecasting methods play a crucial role and provide important guidance for spot trading strategies. However, existing methods primarily analyze and forecast medium- to long-term electricity load by establishing causal relationships between electricity load and economic factors and building vector autoregression models. These methods are not suitable for short-term load forecasting and thus have certain limitations. Summary of the Invention

[0004] This application provides a load forecasting method and related equipment for forecasting short-term loads.

[0005] The first aspect of this application provides a load forecasting method, including:

[0006] Acquire target sample data of the user's historical electricity consumption, the target sample data including multiple target daily electricity consumption data and target time-of-use electricity consumption data corresponding to each target daily electricity consumption data;

[0007] The target daily electricity consumption data is input into a pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily electricity consumption average data.

[0008] Multiple sets of the average daily electricity consumption data of the intermediate typical days are input into a pre-trained linear regression model to obtain the average daily electricity consumption data of the target typical days;

[0009] Clustering regression model is used to cluster the target time-of-use electricity data and the target daily electricity data corresponding to the target time-of-use electricity data to obtain type daily time-of-use weight coefficient data.

[0010] Short-term load is predicted using the target typical daily average electricity consumption data, the typical daily coefficient data, and the type day time-of-use weighting coefficient data.

[0011] Optionally, before obtaining the target sample data of users' historical electricity consumption, the method further includes:

[0012] Obtain initial sample data of the user's historical electricity consumption, the initial sample data including multiple initial time-of-use electricity data;

[0013] Determine whether the initial time-of-use electricity data meets the preset conditions;

[0014] If so, the initial time-of-use power data is input into the initial power correction model to train the initial power correction model;

[0015] The initial time-of-use power data is input into the trained initial power correction model to obtain multiple sets of the target time-of-use power data.

[0016] Calculate the target time-of-use electricity data for any set of target time-of-use electricity data to obtain the target daily electricity data corresponding to any set of target time-of-use electricity data.

[0017] Optionally, before obtaining the target sample data of users' historical electricity consumption, the method further includes:

[0018] Obtain initial sample data of the user's historical electricity consumption, the initial sample data including multiple initial time-of-use electricity data P ki ;

[0019] Determine P ki Is P less than or equal to 0? k(i-1) Is P greater than 0? k(i+1) Is it greater than 0?

[0020] Or, determine P ki Is it greater than The |P (k-n)i ,P (k-1)i | min >0;

[0021] If so, then the P ki Input the initial power correction model P ki =AP k(i-1) +BP k(i+1) In order to train the initial power correction model;

[0022] The target sample data for obtaining users' historical electricity consumption includes:

[0023] P ki Input the trained initial charge correction model P ki =AP k(i-1) +BP k(i+1) To obtain multiple sets of the target time-of-use electricity data;

[0024] Calculate the target time-of-use electricity data for any set of target time-of-use electricity data to obtain the target daily electricity data D corresponding to any set of target time-of-use electricity data. k ;

[0025] Where k is the date, i is the i-th time point, and P ki For the target time-of-use electricity data at the i-th time point on day k, the P k(i-1) For the target time-of-use electricity data at the (i-1)th time point on day k, the P k(i+1) For the target time-of-use electricity data at the (i+1)th time point on day k, the P (k-n)i For the target time-of-use electricity data at the i-th time point on day kn, the P (k-1)i The target time-of-use electricity data is for the i-th time point on day k-1, where A and B are model coefficients of the initial electricity correction model, and D... k The target daily electricity consumption data for day k.

[0026] Optionally, the typical daily load factor regression model includes an initial typical daily average electricity consumption calculation formula, an intermediate typical daily average electricity consumption calculation formula, and a typical daily coefficient calculation formula; the step of inputting the target daily electricity consumption data into the pre-trained typical daily load factor regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily average electricity consumption data includes:

[0027] Select D k D k-1 D k-2 D k-3 D k-4 D k-5 D k-6 As a typical daily electricity consumption array;

[0028] The D of different typical daily electricity consumption arrays k Input the formula for calculating the initial typical daily average electricity consumption. To obtain multiple sets of initial typical daily electricity consumption average data V k ;

[0029] Multiple sets of V kInput the formula for calculating the average daily electricity consumption of the intermediate typical days. To obtain multiple sets of average daily electricity consumption data V for typical intermediate days w ;

[0030] Any group of V k and the V w Input the typical daily coefficient calculation formula The typical daily coefficient data X is obtained. k ;

[0031] Where k is the date, and D is the date. k For the target daily electricity consumption data of day k, the D k-1 For the target daily electricity consumption data of day k-1, the D k-2 For the target daily electricity consumption data of day k-2, the D k-3 For the target daily electricity consumption data of day k-3, the D k-4 For the target daily electricity consumption data of day k-4, the D k-5 For the target daily electricity consumption data of day k-5, the D k-6 For the target daily electricity consumption data of day k-6, the D k-7n For the target daily electricity consumption data of day k-7n, the X k This represents the typical daily coefficient data for day k.

[0032] Optionally, the step of inputting multiple sets of the intermediate typical daily electricity consumption average data into a pre-trained linear regression model to obtain the target typical daily electricity consumption average data includes:

[0033] V w Input the linear regression model V w+1 =M+N0V w +N1V w-1 +N2V w-2 +…+N n V w-n The linear regression model is trained to obtain the target typical daily average electricity consumption data V. w+1 ;

[0034] Wherein, M, N0, N1, N2, N n The model coefficients of the linear regression model are V. w-n For the intermediate typical daily electricity consumption average data of the wn group, the V w+1 This is the average daily electricity consumption data for the target group w+1.

[0035] Optionally, before clustering the target time-of-use electricity data and the target daily electricity data corresponding to the target time-of-use electricity data using a clustering regression model to obtain the type-day time-of-use weight coefficient data, the method further includes:

[0036] The target daily electricity data D of different typical daily electricity arrays k Input the formula for calculating the average daily electricity consumption of the initial typical day. To obtain multiple sets of initial typical daily electricity consumption average data V k ;

[0037] The step of clustering the target time-of-use electricity data and the corresponding target daily electricity data using a clustering regression model to obtain type-day time-of-use weight coefficient data includes:

[0038] Using cluster regression model For any set of target daily electricity data D k-7n and corresponding to the D k-7n The target time-of-use electricity data P (k-7n)i Perform clustering to select the D k-7n The time-of-use power consumption array;

[0039] Through the D k-7n Get the daily average hourly electricity consumption data from the time-of-use electricity consumption array.

[0040] The daily hourly average electricity consumption data of the aforementioned type and the V k Substitute the type of day-to-day weighting coefficient calculation formula Calculations are performed to obtain the daily time-sharing weighting coefficient data X of the aforementioned type. ki ;

[0041] Wherein, n is greater than or equal to 0, k is a date, and D... k For the target daily electricity consumption data of day k, the D k-7n For the target daily electricity consumption data of day k-7n, the P (k-7n)i For the target time-of-use electricity data at the i-th time point on day k-7n, the X ki For the type-day time-sharing weight coefficient data of the i-th time point on the k-th day, the... This represents the average daily hourly electricity consumption data for the i-th time point.

[0042] Optionally, the step of predicting short-term load using the target typical daily average electricity consumption data, the typical daily coefficient data, and the type day time-of-use weighting coefficient data includes:

[0043] The target typical daily electricity consumption average data V w+1and the typical daily coefficient data X k Input daily electricity prediction model D k+7 =X k V w+1 To analyze daily electricity consumption data D in short-term load k+7 Make predictions;

[0044] The daily electricity data D k+7 and the day-to-day weighting coefficient data X of the aforementioned type ki Input time-of-use electricity prediction model P (k+7)i =D k+7 X ki To analyze time-of-use electricity data P in short-term load (k+7)i Make predictions;

[0045] Where k is the date, and V w+1 For the target typical daily electricity consumption average data of group w+1, the X k For typical daily coefficient data of day k, the D k+7 For the target daily electricity consumption data of day k+7, the X ki For the type-day time-sharing weight coefficient data of the i-th time point on the k-th day, the P (k+7)i This is the hourly electricity consumption data for the i-th time point on day k+7.

[0046] A second aspect of this application provides a load forecasting system, comprising:

[0047] The acquisition unit is used to acquire target sample data of the user's historical electricity consumption. The target sample data includes multiple target daily electricity consumption data and target time-of-use electricity consumption data corresponding to each target daily electricity consumption data.

[0048] The input unit is used to input the target daily electricity consumption data into a pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily electricity consumption average data.

[0049] The input unit is also used to input multiple sets of intermediate typical daily electricity consumption average data into a pre-trained linear regression model to obtain target typical daily electricity consumption average data.

[0050] Clustering unit, used to cluster the target time-of-use electricity data and the target daily electricity data corresponding to the target time-of-use electricity data through a clustering regression model, so as to obtain type daily time-of-use weight coefficient data;

[0051] The prediction unit is used to predict short-term load using the target typical daily average electricity consumption data, the typical daily coefficient data, and the type day time-of-use weighting coefficient data.

[0052] Optionally, the system further includes: a judgment unit and a calculation unit;

[0053] The acquisition unit is further configured to acquire initial sample data of the user's historical electricity consumption, the initial sample data including multiple initial time-of-use electricity data;

[0054] The judgment unit is used to determine whether the initial time-of-use electricity data meets the preset conditions;

[0055] The input unit is further configured to input the initial time-of-use power data into the initial power correction model when the initial time-of-use power data meets the preset conditions, so as to train the initial power correction model.

[0056] The input unit is also used to input the initial time-of-use power data into the trained initial power correction model to obtain multiple sets of the target time-of-use power data;

[0057] The calculation unit is used to calculate any set of target time-of-use electricity data to obtain the target daily electricity data corresponding to any set of target time-of-use electricity data.

[0058] Optionally,

[0059] The acquisition unit is further configured to acquire initial sample data of the user's historical electricity consumption, the initial sample data including multiple initial time-of-use electricity data P. ki ;

[0060] The judgment unit is specifically used to judge P. ki Is P less than or equal to 0? k(i-1) Is P greater than 0? k(i+1) Is it greater than 0?

[0061] Alternatively, the judgment unit is specifically used to judge P. ki Is it greater than The |P (k-n)i ,P (k-1)i | min >0;

[0062] If so, the input unit is specifically used to input the P ki Input the initial power correction model P ki =AP k(i-1) +BP k(i+1) In order to train the initial power correction model;

[0063] The input unit is specifically used to input the P ki Input the trained initial charge correction model P ki =AP k(i-1) +BPk(i+1) To obtain multiple sets of the target time-of-use electricity data;

[0064] The calculation unit is specifically used to calculate any set of target time-of-use electricity data to obtain the target daily electricity data D corresponding to any set of target time-of-use electricity data. k ;

[0065] Where k is the date, i is the i-th time point, and P ki For the target time-of-use electricity data at the i-th time point on day k, the P k(i-1) For the target time-of-use electricity data at the (i-1)th time point on day k, the P k(i+1) For the target time-of-use electricity data at the (i+1)th time point on day k, the P (k-n)i For the target time-of-use electricity data at the i-th time point on day kn, the P (k-1)i The target time-of-use electricity data is for the i-th time point on day k-1, where A and B are model coefficients of the initial electricity correction model, and D... k The target daily electricity consumption data for day k.

[0066] Optionally, the system further includes: a selection unit;

[0067] The selection unit is used to select D k D k-1 D k-2 D k-3 D k-4 D k-5 D k-6 As a typical daily electricity consumption array;

[0068] The input unit is used to input the D of different typical daily electricity consumption arrays. k Input the formula for calculating the initial typical daily average electricity consumption. To obtain multiple sets of initial typical daily electricity consumption average data V k ;

[0069] The input unit is specifically used to input multiple sets of V k Input the formula for calculating the average daily electricity consumption of the intermediate typical days. To obtain multiple sets of average daily electricity consumption data V for typical intermediate days w ;

[0070] The input unit is specifically used to input any set of V k and the V w Input the typical daily coefficient calculation formula The typical daily coefficient data X is obtained. k ;

[0071] Where k is the date, and D is the date. k For the target daily electricity consumption data of day k, the D k-1 For the target daily electricity consumption data of day k-1, the D k-2 For the target daily electricity consumption data of day k-2, the D k-3 For the target daily electricity consumption data of day k-3, the D k-4 For the target daily electricity consumption data of day k-4, the D k-5 For the target daily electricity consumption data of day k-5, the D k-6 For the target daily electricity consumption data of day k-6, the D k-7n For the target daily electricity consumption data of day k-7n, the X k This represents the typical daily coefficient data for day k.

[0072] Optionally,

[0073] The input unit is specifically used to process multiple sets of intermediate typical daily electricity consumption average data V w Input the linear regression model V w+1 =M+N0V w +N1V w-1 +N2V w-2 +…+N n V w-n The linear regression model is trained to obtain the target typical daily average electricity consumption data V. w+1 ;

[0074] Wherein, M, N0, N1, N2, N n The model coefficients of the linear regression model are V. w-n For the intermediate typical daily electricity consumption average data of the wn group, the V w+1 This is the average daily electricity consumption data for the target group w+1.

[0075] Optionally,

[0076] The input unit is specifically used to input the target daily electricity data D from different typical daily electricity arrays. k Input the formula for calculating the average daily electricity consumption of the initial typical day. To obtain multiple sets of initial typical daily electricity consumption average data V k ;

[0077] The clustering unit is specifically used to perform clustering regression model For any set of target daily electricity data D k-7n and corresponding to the D k-7n The target time-of-use electricity data P (k-7n)i Perform clustering to select the Dk-7n The time-of-use power consumption array;

[0078] The acquisition unit is specifically used to obtain the D k-7n Get the daily average hourly electricity consumption data from the time-of-use electricity consumption array.

[0079] The calculation unit is specifically used to process the daily hourly average electricity consumption data of the aforementioned type. and the V k Substitute the type of day-to-day weighting coefficient calculation formula Calculations are performed to obtain the daily time-sharing weighting coefficient data X of the aforementioned type. ki ;

[0080] Wherein, n is greater than or equal to 0, k is a date, and D... k For the target daily electricity consumption data of day k, the D k-7n For the target daily electricity consumption data of day k-7n, the P (k-7n)i For the target time-of-use electricity data at the i-th time point on day k-7n, the X ki For the type-day time-sharing weight coefficient data of the i-th time point on the k-th day, the... This represents the average daily hourly electricity consumption data for the i-th time point.

[0081] Optionally,

[0082] The input unit is specifically used to input the target typical daily average electricity consumption data V. w+1 and the typical daily coefficient data X k Input daily electricity prediction model D k+7 =X k V w+1 To analyze daily electricity consumption data D in short-term load k+7 Make predictions;

[0083] The input unit is specifically used to input the daily electricity data D k+7 and the day-to-day weighting coefficient data X of the aforementioned type ki Input time-of-use electricity prediction model P (k+7)i =D k+7 X ki To analyze time-of-use electricity data P in short-term load (k+7)i Make predictions;

[0084] Where k is the date, and V w+1 For the target typical daily electricity consumption average data of group w+1, the X k For typical daily coefficient data of day k, the D k+7 For the target daily electricity consumption data of day k+7, the X kiFor the type-day time-sharing weight coefficient data of the i-th time point on the k-th day, the P (k+7)i This is the hourly electricity consumption data for the i-th time point on day k+7.

[0085] The second aspect of the embodiments of this application provides a method for performing the method described in the first aspect.

[0086] A third aspect of this application provides a load forecasting apparatus, comprising:

[0087] Central processing unit, memory, input / output interfaces, wired or wireless network interfaces, and power supply;

[0088] The memory is either a short-term storage memory or a persistent storage memory;

[0089] The central processing unit is configured to communicate with the memory and execute instructions in the memory to perform the method described in the first aspect.

[0090] A fourth aspect of this application provides a computer-readable storage medium, characterized in that the computer-readable storage medium includes instructions that, when executed on a computer, cause the computer to perform the method described in the first aspect.

[0091] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0092] The load forecasting method proposed in this application first obtains target sample data of users' historical electricity consumption. This target sample data includes multiple target daily electricity consumption data and target time-of-use electricity consumption data corresponding to each target daily electricity consumption data. Then, the target daily electricity consumption data is input into a pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily average electricity consumption data. Next, the multiple sets of intermediate typical daily average electricity consumption data are input into a pre-trained linear regression model to obtain the target typical daily average electricity consumption data. Subsequently, a clustering regression model is used to cluster the target time-of-use electricity consumption data and the target daily electricity consumption data corresponding to the target time-of-use electricity consumption data to obtain type-day time-of-use weight coefficient data. Finally, short-term load is predicted using the target typical daily average electricity consumption data, typical daily coefficient data, and type-day time-of-use weight coefficient data. In summary, this application embodiment, by combining the characteristics of the electricity market and employing multiple models for short-term forecasting of user electricity consumption data, can effectively avoid the limitations of single-model forecasting, thereby improving forecast accuracy and effectively assisting electricity sales companies in spot trading. Attached Figure Description

[0093] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0094] Figure 1 This is a schematic flowchart of a load forecasting method disclosed in an embodiment of this application;

[0095] Figure 2 This is a schematic flowchart of another load forecasting method disclosed in an embodiment of this application;

[0096] Figure 3 This is a time-sharing curve analysis diagram disclosed in the embodiments of this application;

[0097] Figure 4 This is a comparison chart of time-sharing curves disclosed in the embodiments of this application;

[0098] Figure 5 This is a time-of-use power consumption calculation diagram disclosed in an embodiment of this application;

[0099] Figure 6 This is a schematic diagram of the structure of a load forecasting system disclosed in an embodiment of this application;

[0100] Figure 7 This is a schematic diagram of the structure of a load prediction device disclosed in an embodiment of this application. Detailed Implementation

[0101] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0102] It should be noted that the embodiments of this application mainly address the technical problems of day-ahead load forecasting in the current electricity spot market. To elaborate, the day-ahead market refers to the market where, on the day before the operating date, generator unit bids are combined with the forecast of centrally dispatched load. The SCUC algorithm is used to first list the operating sequence of all units in the market, and then the SCED algorithm, considering boundary information, is used to finally clear the day-ahead price for the entire market. The real-time market, on the other hand, uses the previous day's unit operating and bidding information, combined with the latest boundary information on the operating day, to clear the real-time price for the entire market according to the SCED algorithm.

[0103] Please see Figure 1 , Figure 1This is a flowchart illustrating a load forecasting method disclosed in an embodiment of this application. It includes steps 101-105.

[0104] 101. Obtain target sample data of users' historical electricity consumption.

[0105] Before forecasting short-term power load, it is necessary to obtain target sample data of users' historical electricity consumption. It should be noted that this target sample data includes target daily electricity consumption data and the corresponding target time-of-use electricity consumption data. Specifically, the target daily electricity consumption data is the cumulative 24-hour electricity consumption of each user, i.e., P = P0 + P1 + ... + P 22 +P 23 Correspondingly, P represents the cumulative electricity consumption over the 24 hours of the day, P0, P1...P 22 and P 23 This refers to the target hourly electricity consumption data for 0:00, 1:00, ..., 22:00 and 23:00 on the same day.

[0106] Since the target sample data consists of daily electricity consumption data over multiple days, it includes multiple target daily electricity consumption data and target hourly electricity consumption data corresponding to each target daily electricity consumption data.

[0107] It should be noted that, for ease of description, the target daily electricity consumption data will be described using the actual daily electricity consumption data, and the corresponding target hourly electricity consumption data will be described using the actual hourly electricity consumption data. This will not be elaborated on further.

[0108] 102. Input the target daily electricity consumption data into the pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily electricity consumption average data.

[0109] Once the target day's electricity consumption data is obtained, it can be used as the actual daily electricity consumption for the electricity sales company's users. Then, the target day's electricity consumption data within the next week is selected as the typical day, that is, the target day's electricity consumption data from Monday to Sunday is used as the typical day's electricity consumption array.

[0110] Then, the obtained target daily electricity consumption data can be input into the typical daily load factor regression model according to different typical daily electricity consumption arrays. For example, using May 23rd as the baseline, the target daily electricity consumption data for all Mondays in May can be input into the typical daily load factor regression model, that is, the target daily electricity consumption data for May 2nd, May 9th, May 16th, and May 23rd can be input into the typical daily load factor regression model to obtain the initial typical daily electricity consumption average data for May 23rd. Correspondingly, using other dates as the baseline, the initial typical daily electricity consumption average data for Tuesdays to Sundays associated with May 23rd can also be obtained.

[0111] Then, the average of the initial typical daily electricity consumption data from Monday to Sunday corresponding to Monday, May 23, is calculated to obtain a set of intermediate typical daily electricity consumption average data. Correspondingly, if the date changes, multiple sets of intermediate typical daily electricity consumption average data can be obtained.

[0112] Dividing the initial average daily electricity consumption data by the intermediate average daily electricity consumption data yields the typical daily coefficient data corresponding to May 23.

[0113] It should be noted that the above description of the date May 23 is only one specific example. The results of the data acquisition may be different for different dates. For the sake of simplicity, this will not be elaborated on further.

[0114] 103. Input multiple sets of intermediate typical daily electricity consumption average data into a pre-trained linear regression model to obtain target typical daily electricity consumption average data.

[0115] In step 102, since multiple sets of intermediate typical daily electricity consumption average data have been obtained, these data can be input into the pre-trained linear regression model to obtain the target typical daily electricity consumption average data.

[0116] Correspondingly, the average daily electricity consumption data for this target was obtained through linear regression of multiple recent sets of average daily electricity consumption data.

[0117] 104. Cluster the target time-of-use electricity data and the target daily electricity data corresponding to the target fractional electricity data using a clustering regression model to obtain the type daily time-of-use weight coefficient data.

[0118] Since the target sample data of the user's historical electricity consumption was obtained in step 101, and the target sample data includes the target daily electricity consumption data for different dates and the target hourly electricity consumption data corresponding to each target daily electricity consumption data, the target hourly electricity consumption data and the target daily electricity consumption data corresponding to the target hourly electricity consumption data can be clustered to obtain the type day hourly weight coefficient data.

[0119] Specifically, multiple target hourly electricity consumption data points for the same day can be divided by the target daily electricity consumption data to cluster the resulting curves. Then, by selecting the hourly electricity consumption array for that day from the curves, the average electricity consumption for the same hour within the same array can be obtained. For example, if the day is May 23rd, a Monday, and the hourly data is selected at 0:00, the average hourly electricity consumption data for Mondays in May at 0:00 can be obtained.

[0120] Then, the average value can be divided by the average value of the initial typical daily electricity consumption corresponding to May 23 obtained in step 102, so as to obtain the type daily time-of-day weight coefficient data at 0:00 on May 23.

[0121] It should be noted that the above description of the date May 23 is only one specific example. The results of the data calculation may be different for different dates and different time periods. For the sake of simplicity, this will not be elaborated on further.

[0122] 105. Short-term load is predicted using the average daily electricity consumption data, typical daily coefficient data, and time-of-use weighting coefficient data of the target day.

[0123] By obtaining the target typical daily electricity average data and typical daily coefficient data in steps 101-104, the predicted daily electricity data can be obtained, thereby predicting the short-term load.

[0124] Then, by combining the predicted daily electricity consumption data with the time-of-use weighting coefficient data, the predicted daily time-of-use electricity consumption data can be obtained, thereby predicting the time-of-use electricity consumption of short-term load.

[0125] Specifically, taking Monday, May 23 as an example, based on the average daily electricity consumption data and typical daily coefficient data corresponding to Monday, May 23, the predicted daily electricity consumption data for Monday, May 30 can be obtained. In other words, the electricity consumption for the following Monday can be predicted.

[0126] Once the predicted daily electricity consumption data for Monday, May 30th is obtained, the hourly electricity consumption at 0:00 on Monday, May 30th can be predicted based on the time-of-use weighting coefficient data for the type day at 0:00 on Monday, May 23rd.

[0127] In summary, this allows for the prediction of short-term power load.

[0128] The load forecasting method proposed in this application first obtains target sample data of users' historical electricity consumption. This target sample data includes multiple target daily electricity consumption data and target time-of-use electricity consumption data corresponding to each target daily electricity consumption data. Then, the target daily electricity consumption data is input into a pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily average electricity consumption data. Next, the multiple sets of intermediate typical daily average electricity consumption data are input into a pre-trained linear regression model to obtain the target typical daily average electricity consumption data. Subsequently, a clustering regression model is used to cluster the target time-of-use electricity consumption data and the target daily electricity consumption data corresponding to the target time-of-use electricity consumption data to obtain type-day time-of-use weight coefficient data. Finally, short-term load is predicted using the target typical daily average electricity consumption data, typical daily coefficient data, and type-day time-of-use weight coefficient data. In summary, this application embodiment, by combining the characteristics of the electricity market and employing multiple models for short-term forecasting of user electricity consumption data, can effectively avoid the limitations of single-model forecasting, thereby improving forecast accuracy and effectively assisting electricity sales companies in spot trading.

[0129] The following is a detailed description of a load forecasting method proposed in an embodiment of this application. Please refer to [link / reference]. Figure 2 , Figure 2 This is a flowchart illustrating another load forecasting method disclosed in this application. It includes steps 201-208. For ease of description, this embodiment uses electricity consumption data from May 1, 2022 to May 26, 2022, published by the Guangdong Power Exchange Center system for detailed explanation.

[0130] 201. Obtain initial sample data of users' historical electricity consumption.

[0131] Before forecasting short-term electricity load, it is necessary to obtain initial sample data of users' historical electricity consumption. It should be noted that this initial sample data includes multiple initial time-of-use (TOU) electricity data points. Correspondingly, the initial daily electricity data is the cumulative 24-hour electricity consumption of users on that day, i.e., P = P0 + P1 + ... + P 22 +P 23 Correspondingly, P represents the cumulative electricity consumption over the 24 hours of the day, P0, P1...P 22 and P 23 This refers to the initial hourly electricity consumption data for 0:00, 1:00, ..., 22:00 and 23:00 on the corresponding day.

[0132] 202. Determine whether the initial time-of-use electricity data meets the preset conditions. If yes, proceed to step 203.

[0133] After obtaining the initial time-of-use electricity data, since the electricity data is published by the Guangdong Power Exchange Center system, there may be instances where the published electricity data is less than 0 (abnormal data). Therefore, it is necessary to correct the historical electricity consumption data (basic data). Specifically, this means determining whether the initial time-of-use electricity data meets the preset conditions. If so, proceed to step 203.

[0134] Specifically, there are two ways to determine this: if the initial time-of-use electricity data P... ki One of the following conditions must be met:

[0135] The first type: P ki Is P less than or equal to 0? k(i-1) Is P greater than 0? k(i+1) Is it greater than 0? Correspondingly, if P ki P k(i-1) and P k(i+1) If the above judgment conditions are met, then P needs to be... ki Make corrections.

[0136] The second method: Determine P ki Is it greater than C|P? (k-n)i ,P (k-1)i | max .in, And |P (k-n)i ,P (k-1)i | min >0.

[0137] It should be noted that, where k is the date, i is the i-th time point, and P... ki For the target time-of-use electricity data at the i-th time point on day k, P k(i-1) For the target time-of-use electricity data at the (i-1)th time point on day k, P k(i+1) For the target time-of-use electricity data at the (i+1)th time point on day k, P (k-n)i For the target time-of-use electricity data at the i-th time point on day kn, P (k-1)i For the target time-of-use electricity data at the i-th time point on day k-1, the corresponding P (k-n)i This refers to the electricity consumption at the nth time point on the nth day prior to the date. For example, if the Guangdong Power Exchange Center releases the historical time-of-use electricity consumption data for May 26th on May 30th, then the electricity sales company can obtain the time-of-use electricity consumption data for k-4. The data for May 25th and 24th are represented by k-5 and k-6, respectively.

[0138] It should also be noted that the above description of the date May 30th is only one specific example. The results of the data calculation may be different for different dates and different time periods. For the sake of simplicity, this will not be elaborated on further.

[0139] 203. Input the initial time-of-use electricity data into the initial electricity correction model to train the initial electricity correction model.

[0140] If P ki If any of the above judgment conditions are met, then P will be... ki Input initial charge correction model P ki =AP k(i-1) +BP k(i+1) This is used to train the initial power correction model. It should be noted that A and B are values ​​that need to be repeatedly corrected by the model. In other words, A and B at this point can be considered unknown values.

[0141] 204. Input the initial time-of-use electricity data into the trained initial electricity correction model to obtain multiple sets of target time-of-use electricity data and target daily electricity data.

[0142] After inputting the initial time-of-use electricity data into the trained initial electricity correction model, the target daily electricity data D can be obtained. k and corresponding to D k Different time-sharing target time-sharing electricity data P ki It's not hard to understand that D k =P k0 +P k1 +…+P k22 +P k23 .

[0143] Specifically, the initial power correction model refers to the power data anomaly correction tool, taking the correction on the same day as an example:

[0144] May 16th was a Monday. At 1:00 AM, the electricity volume released by the Guangdong Electric Power Exchange Center was -100,000 kWh; at 12:00 AM, it was 1,300,000 kWh; and at 2:00 AM, it was 1,500,000 kWh. On May 9th, also a Monday, the electricity volume released by the Guangdong Electric Power Exchange Center at 1:00 AM was 1,600,000 kWh; at 12:00 AM, it was 1,400,000 kWh; and at 2:00 AM, it was 1,800,000 kWh. By comparison, the data at 1:00 AM on the 16th can be determined to be abnormal. (This information was obtained through P...) ki =AP k(i-1) +BP k(i+1) Linear regression yielded A = B = 0.5, and the electricity consumption data at 1:00 AM on the 16th was corrected to 1.4 million kilowatt-hours. Training can select the start and end times of the data for a one-time training session; for example, selecting electricity consumption data from May 1st to 23rd.

[0145] 205. Input the target daily electricity consumption data into the pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data, initial typical daily average electricity consumption data, and multiple sets of intermediate typical daily average electricity consumption data corresponding to the target daily electricity consumption data.

[0146] It should be noted that the typical daily load factor regression model includes the formula for calculating the initial average typical daily electricity consumption, the formula for calculating the average intermediate typical daily electricity consumption, and the formula for calculating the typical daily load factor.

[0147] First, select D. k D k-1 D k-2 D k-3 D k-4 D k-5 D k-6 As a typical daily (Monday to Sunday) electricity consumption array, for example, the electricity consumption data from May 16th to 29th is selected as the electricity consumption array, with the 16th and 23rd corresponding to Mondays, and the 22nd and 29th corresponding to Sundays.

[0148] For details, please refer to Table 1, which shows the specific values ​​of electricity consumption for one typical day and its hourly rate.

[0149]

[0150]

[0151] Set initial typical daily average electricity consumption data V k To be with D k For the same typical daily average electricity consumption, D represents different typical daily electricity consumption arrays. k Input the formula for calculating the average daily electricity consumption of the initial typical day. To obtain multiple sets of V k Therefore, V1 represents the average electricity consumption on the 16th and 23rd, and V2 represents the average electricity consumption on the 17th and 24th.

[0152] For details, please refer to Table 2, which shows one type of initial typical daily electricity consumption average data V. k The specific value.

[0153] typical day Total power consumption Daily average electricity consumption Monday 10534.53 3511.51 Tuesday 11020.98 3673.66 Wednesday 10937.88 3645.96 Thursday 14696.71 3674.18 Friday 11251.56 3750.52 Saturday 10972.41 3657.47 Sunday 9795.26 3265.09

[0154] Set the average daily electricity consumption data V (for intermediate typical days) w As a set of typical daily electricity consumption averages, multiple sets of V k Input the formula for calculating the average daily electricity consumption of a typical intermediate day. To obtain multiple sets of intermediate typical daily electricity consumption average data V w .

[0155] As shown in Table 2, V at this time w The value is the average daily electricity consumption on a typical day (Monday to Sunday), which is 3680.36.

[0156] Finally, set the typical daily coefficient data X. k D kThe typical daily coefficient, for any set of V k and V w Input the typical daily coefficient calculation formula This will allow us to obtain typical daily coefficient data X. k .

[0157] For details, please refer to Table 3, which shows one typical daily coefficient X. k The specific value.

[0158] typical day Typical daily coefficient Monday 1 0.9541 Tuesday 2 0.9982 Wednesday 3 0.9907 Thursday 4 0.9983 Friday 5 1.0191 Saturday 6 0.9938 Sunday 7 0.8872

[0159] It should be noted that k represents the date, and D... k For the target daily electricity consumption data on day k, D k-1 For the target daily electricity consumption data of day k-1, D k-2 For the target daily electricity consumption data of day k-2, D k-3 For the target daily electricity consumption data of day k-3, D k-4 For the target daily electricity consumption data of day k-4, D k-5 For the target daily electricity consumption data of day k-5, D k-6 For the target daily electricity consumption data of day k-6, D k-7n For the target daily electricity consumption data on day k-7n, X k This represents the typical daily coefficient data for day k. For ease of description, the specific meanings of each expression and symbol will not be elaborated further.

[0160] It should also be noted that Tables 1, 2 and 3 are just one specific implementation of the process of calculating the typical daily coefficient. When the time-of-use electricity data is different, the corresponding typical daily coefficient will also be different, which will not be elaborated here.

[0161] 206. Input multiple sets of intermediate typical daily electricity consumption average data into a pre-trained linear regression model to obtain target typical daily electricity consumption average data.

[0162] Once multiple sets of average daily electricity consumption data for typical intermediate days are obtained, the average daily electricity consumption data for typical intermediate days can be V. w Input linear regression model V w+1 =M+N0V w +N1V w-1 +N2V w-2 +…+N n V w-n This is used to train a linear regression model and obtain the target typical daily average electricity consumption data V. w+1 .

[0163] Where M, N0, N1, N2, ..., N nThese are the model coefficients for the linear regression model, generally calculated from linear regression. Influencing factors typically include temperature, humidity, etc. However, the specific content of the influencing factors is not limited here, and will not be elaborated further. V w-n V represents the average daily electricity consumption data of the intermediate typical group wn. w+1 This is the average daily electricity consumption data for the target group w+1.

[0164] Furthermore, during training, the start and end times of the data can be selected for one-time training, for example, selecting power data from May 1st to 23rd.

[0165] For details, please refer to Table 4, which shows the average daily electricity consumption data (V) for one type of target. w+1 The specific value.

[0166] date typical day coefficient Actual power consumption Coefficient prediction Predict the final value Workday Forecast Note (Temperature) 2022-05-16 Monday 0.9635 3254.7070 3275.9027 3488.00 3400 15-23 2022-05-17 Tuesday 0.9978 3411.6701 3422.6141 3700.00 3430 19-29 2022-05-18 Wednesday 0.9970 3488.4503 3489.3528 3768.00 3500 20-30 2022-05-19 Thursday 0.9960 3622.4187 3620.4951 3700.00 3635 24-31 2022-05-20 Friday 1.0159 3800.6847 3809.5243 3730.00 3750 25-32 2022-05-21 Saturday 0.9905 3697.9519 3694.6790 3670.00 3730 22-29 2022-05-22 Sunday 0.9208 3342.9689 3361.0540 3300.00 3650 22-27 2022-05-23 Monday 0.9635 3637.4818 3613.1280 3420.00 3750 24-29 2022-05-24 Tuesday 0.9978 3800.1902 3791.8174 3642.00 3800 25-31 2022-05-25 Wednesday 0.9970 3799.0426 3788.4402 3700.00 3800 25-31 2022-05-26 Thursday 0.9960 3803.8657 3784.8367 3730.00 3800 25-31

[0167] Among them, the weekday forecast is the average daily electricity consumption data V of the target typical day. w+1 If the actual battery consumption has already been published, then a regression analysis will be performed.

[0168] It should also be noted that Table 4 shows the average daily electricity consumption data V for calculating the target. w+1 In one specific embodiment, when the power data is different, the generated V w+1 The corresponding ones will also be different, but the details will not be elaborated here.

[0169] 207. Cluster the target time-of-use electricity data, the target daily electricity data corresponding to the target time-of-use electricity data, and the initial typical daily average electricity data using a clustering regression model to obtain the type daily time-of-use weight coefficient.

[0170] As can be seen from step 204, P k0 ,P k1 ,...,P k22 ,P k23 D k 24-hour time-of-use power consumption.

[0171] Using cluster regression model For any set of target daily electricity data D k-7n Perform clustering to select the D k-7n The time-of-use (TOU) power consumption array. Specifically, through... Clustering is performed on the curves, and D is selected by approximating the curves. k-7n The time-of-use electricity array is used to obtain the daily time-of-use average electricity data.

[0172] Then, set the type daily time-sharing weight coefficient data X.ki D k The time-sharing coefficient at time i, combined with V obtained in step 205. k X can then be calculated. ki Correspondingly, It should be noted that, for example, when i = 0, The corresponding ones are, It should also be noted that n is greater than or equal to 0, k is the date, and D k For the target daily electricity consumption data on day k, D k-7n For the target daily electricity consumption data on day k-7n, P (k-7n)i X represents the target time-of-use electricity data for the i-th time point on day k-7n. ki This refers to the time-sharing weight coefficient data for the type of day at the i-th time point on day k. This represents the average daily hourly electricity consumption data for the i-th time point. It refers to the ratio of the hourly electricity consumption at the i-th time point on day k-7n to the total electricity consumption on day k-7n.

[0173] It should be added that, in one embodiment, the base data is selected from May 2nd to 29th, resulting in 4 complete sets of weekly data. The 2nd, 9th, 16th, and 23rd are all Mondays. Four sets of Monday intraday curves are calculated. The similarity of the four curves is determined, and curves with significant differences are discarded (for example, the curve for the 2nd is significantly different from the curves for the 9th, 16th, and 23rd). The average electricity consumption at midnight on the 9th, 16th, and 23rd corresponds to the average electricity consumption at midnight on Monday. The average electricity consumption at 0:00 on the 10th, 17th, and 24th corresponds to the average electricity consumption at 0:00 on Tuesday.

[0174] For details, please refer to Table 5, which shows the average daily hourly electricity consumption data for one type. The specific value.

[0175]

[0176]

[0177] The average daily electricity consumption is V obtained in step 205. k The corresponding times are 0:00, 1:00, 2:00, and 3:00. The corresponding type of daily hourly average electricity consumption data.

[0178] For details, please refer to Table 6, which shows the daily time-sharing weighting coefficient data for one type. ki The specific value.

[0179] typical day Typical daily coefficient 0 o'clock 1 o'clock 2 o'clock 3 o'clock Monday 1 0.9541 3.7878％ 3.7480％ 3.6244％ 3.5689％ Tuesday 2 0.9982 4.0240％ 3.9792％ 3.8579％ 3.8200％ Wednesday 3 0.9907 4.0996％ 4.0112％ 3.9045％ 3.8511％ Thursday 4 0.9983 3.9744％ 3.9856％ 3.8540％ 3.7876％ Friday 5 1.0191 4.0330％ 3.9737％ 3.8691％ 3.8244％ Saturday 6 0.9938 4.1537％ 4.1364％ 4.0122％ 3.9346％ Sunday 7 0.8872 4.4298％ 4.4120％ 4.3047％ 4.2096％

[0180] Among them, the typical daily coefficient is the typical daily coefficient data X obtained in step 205. k 0:00, 1:00, 2:00, and 3:00 correspond to the time-sharing weighting coefficient data X for the type day. k0 X k1 X k2 X k3 .

[0181] It should also be noted that Tables 5 and 6 are for calculating the coefficient data X of typical days. k and type of daily hourly average electricity consumption data One specific embodiment is that when the power data is different, the generated X k and The corresponding ones will also be different, but the details will not be elaborated here.

[0182] 208. Short-term load is predicted using the average daily electricity consumption data, typical daily coefficient data, and time-of-use weighting coefficients for different types of days.

[0183] V, the average daily electricity consumption data of the target w+1 and typical daily coefficient data X k Input daily electricity prediction model D k+7 =X k V w+1 To analyze daily electricity consumption data D in short-term load k+7 Make predictions;

[0184] Daily electricity data D k+7 and type daily time-sharing weight coefficient data X ki Input time-of-use electricity prediction model P (k+7)i =D k+7 X ki To analyze time-of-use electricity data P in short-term load (k+7)i Make predictions.

[0185] For example, when i = 0, P can be calculated. (k+7)0 Time-of-use electricity, P (k+7)0 =D k+7 X k0 Similarly, P can be calculated. (k+7)0 P (k+7)1 , ..., P (k+7)23 .

[0186] Where k is the date, V w+1 X represents the average daily electricity consumption data for the target typical day in group w+1. k D represents the typical daily coefficient data for day k. k+7 For the target daily electricity consumption data on day k+7, X kiFor the type-day time-sharing weight coefficient data of the i-th time point on the k-th day, P (k+7)i This is the hourly electricity consumption data for the i-th time point on day k+7.

[0187] For example, assuming V2 is the average daily electricity consumption for the next week, then

[0188] V1=M+N0V0+N1V -1 +…+N n V -n ,

[0189] The predicted electricity consumption for the next 1-7 days are D1, D2, D3, ..., D7.

[0190] X1, X2, X3, ..., X4 represent the typical daily coefficients for the next seven days, then D1 = X1V1.

[0191] X 1，0 Let P represent the coefficient of D1 at point 0, then the charge P of D1 at point 0. 1,0 =D1X 1,0 .

[0192] For details, please refer to Table 7, which shows one method for predicting daily electricity consumption D. k+7 The specific value.

[0193] date typical day coefficient Coefficient prediction Predict the final value Workday Forecast Temperature 2022-05-27 Friday 1.0159 3860.3179 3800.00 3800 24-30 2022-05-28 Saturday 0.9905 3764.0161 3730.00 3800 25-31 2022-05-29 Sunday 0.9208 3499.1795 3480.00 3800 25-32 2022-05-30 Monday 0.9635 3680.5730 3680.00 3820 25-33 2022-05-31 Tuesday 0.9959 3814.2625 3810.00 3830 25-32 2022-05-27 Friday 1.0159 3860.3179 3800.00 3800 24-30

[0194] For details, please refer to Table 8, which shows one method for predicting daily time-of-use electricity consumption P. (k+7)i The specific value.

[0195] date typical day Predicted power consumption 0 o'clock 1 o'clock 2 o'clock 3 o'clock 2022-05-27 Friday 3800.00 153.25 151.00 147.03 145.33 2022-05-28 Saturday 3730.00 154.93 154.29 149.65 146.76 2022-05-29 Sunday 3480.00 154.16 153.54 149.80 146.49 2022-05-30 Monday 3680.00 139.39 137.93 133.38 131.34 2022-05-31 Tuesday 3810.00 153.32 151.61 146.99 145.54

[0196] It should also be noted that Tables 7 and 8 show the predicted daily electricity consumption D. k+7 and predicted daily hourly electricity consumption P (k+7)i One specific embodiment is that when the power data is different, the generated D k+7 and P (k+7)i The corresponding ones will also be different, but the details will not be elaborated here.

[0197] The load forecasting method proposed in this embodiment addresses the shortcomings of existing medium- and long-term power load forecasting methods by inventing a short-term power load forecasting method based on multiple forecasting models. This solves the problem of large deviations in existing analyses and improves the feasibility of the proposed method.

[0198] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0199] If the plan involves sensitive information (such as user information or corporate information), it should state that the collection, use, and processing of sensitive information must comply with the laws, regulations, and standards of the relevant countries and regions, and must be carried out with the permission or consent of the relevant entities (such as users or enterprises).

[0200] Please refer to Figure 3 , Figure 4 , Figure 5 , Figure 3 This is a time-sharing curve analysis diagram disclosed in the embodiments of this application; Figure 4 This is a comparison chart of time-sharing curves disclosed in the embodiments of this application; Figure 5 This is a time-of-use power consumption calculation diagram disclosed in an embodiment of this application.

[0201] in, Figure 3 For X in a time-of-use electricity array in step 207 ki A curve analysis of the mean. Figure 4 In step 207, the daily average hourly electricity consumption data is stored in an hourly electricity consumption array. The curve analysis graph. Figure 5 This is a chart showing the proportion of daily hourly electricity consumption to total daily electricity consumption.

[0202] The above describes an embodiment of a load forecasting method according to this application. The following describes an embodiment of a load forecasting system according to this application. Please refer to [link / reference]. Figure 6 , Figure 6 This is a schematic diagram of the structure of a load forecasting system disclosed in an embodiment of this application.

[0203] The acquisition unit 601 is used to acquire target sample data of the user's historical electricity consumption. The target sample data includes multiple target daily electricity consumption data and target time-of-use electricity consumption data corresponding to each target daily electricity consumption data.

[0204] The input unit 602 is used to input the target daily electricity consumption data into the pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily electricity consumption average data.

[0205] The input unit 602 is also used to input multiple sets of intermediate typical daily electricity consumption average data into a pre-trained linear regression model to obtain target typical daily electricity consumption average data;

[0206] Clustering unit 603 is used to cluster the target time-of-use electricity data and the target daily electricity data corresponding to the target time-of-use electricity data through a clustering regression model to obtain the type daily time-of-use weight coefficient data.

[0207] The prediction unit 604 is used to predict short-term load using the target typical daily average electricity consumption data, typical daily coefficient data, and type day time-of-use weighting coefficient data.

[0208] Optionally, the system further includes: a judgment unit 605 and a calculation unit 606;

[0209] The acquisition unit 601 is also used to acquire initial sample data of the user's historical electricity consumption, which includes multiple initial time-of-use electricity data.

[0210] The judgment unit 605 is used to determine whether the initial time-sharing power data meets the preset conditions.

[0211] The input unit 602 is also used to input the initial time-of-use power data into the initial power correction model when the initial time-of-use power data meets the preset conditions, so as to train the initial power correction model.

[0212] The input unit 602 is also used to input the initial time-of-use power data into the trained initial power correction model in order to obtain multiple sets of target time-of-use power data.

[0213] The calculation unit 606 is used to calculate any set of target time-of-use electricity data to obtain target daily electricity data corresponding to any set of target time-of-use electricity data.

[0214] Optionally,

[0215] The acquisition unit is also used to acquire initial sample data of the user's historical electricity consumption, which includes multiple initial time-of-use electricity data P. ki ;

[0216] Judgment unit 605 is specifically used to determine P ki Is P less than or equal to 0? k(i-1) Is P greater than 0? k(i+1) Is it greater than 0?

[0217] Alternatively, judgment unit 605 is specifically used to judge P. ki Is it greater than |P (k-n)i ,P (k-1)i | min >0;

[0218] If so, input unit 602 is specifically used to input P ki Input initial charge correction model P ki =AP k(i-1) +BP k(i+1) This is to train the initial power correction model;

[0219] Input unit 602 is specifically used to input P ki Input the trained initial charge correction model P ki =AP k(i-1) +BP k(i+1) To obtain multiple sets of target time-of-use electricity data;

[0220] The calculation unit 606 is specifically used to calculate any set of target time-of-use electricity data to obtain the target daily electricity data D corresponding to any set of target time-of-use electricity data. k ;

[0221] Where k is the date, i is the i-th time point, and P ki For the target time-of-use electricity data at the i-th time point on day k, P k(i-1) For the target time-of-use electricity data at the (i-1)th time point on day k, P k(i+1) For the target time-of-use electricity data at the (i+1)th time point on day k, P (k-n)i For the target time-of-use electricity data at the i-th time point on day kn, P (k-1)i Let A and B be the target time-of-use electricity data for the i-th time point on day k-1, and let D be the model coefficients of the initial electricity correction model. k This represents the target daily electricity consumption data for day k.

[0222] Optionally, the system further includes: a selection unit 607;

[0223] Select unit 607, used to select D k D k-1 D k-2 D k-3 D k-4 D k-5 D k-6 As a typical daily electricity consumption array;

[0224] Input unit 602 is used to input D of different typical daily electricity consumption arrays. k Input the formula for calculating the average daily electricity consumption of the initial typical day. To obtain multiple sets of initial typical daily electricity consumption average data V k ;

[0225] Input unit 602 is specifically used to input multiple sets of V k Input the formula for calculating the average daily electricity consumption of a typical intermediate day. To obtain multiple sets of intermediate typical daily electricity consumption average data V w ;

[0226] Input unit 602 is specifically used to input any set of V k and V w Input the typical daily coefficient calculation formula Obtain typical daily coefficient data X k ;

[0227] Where k is the date, D k For the target daily electricity consumption data on day k, D k-1 For the target daily electricity consumption data of day k-1, D k-2 For the target daily electricity consumption data of day k-2, D k-3 For the target daily electricity consumption data of day k-3, D k-4 For the target daily electricity consumption data of day k-4, D k-5 For the target daily electricity consumption data of day k-5, D k-6 For the target daily electricity consumption data of day k-6, D k-7n For the target daily electricity consumption data on day k-7n, X k This represents the typical daily coefficient data for day k.

[0228] Optionally,

[0229] Input unit 602 is specifically used to process multiple sets of intermediate typical daily electricity consumption average data V w Input linear regression model V w+1 =M+N0V w +N1V w-1 +N2V w-2 +…+N n V w-n This is used to train a linear regression model and obtain the target typical daily average electricity consumption data V. w+1 ;

[0230] Where M, N0, N1, N2, N n V represents the model coefficients of the linear regression model. w-n V represents the average daily electricity consumption data of the intermediate typical group wn. w+1 This is the average daily electricity consumption data for the target group w+1.

[0231] Optionally,

[0232] Input unit 602 is specifically used to input the target daily electricity data D of different typical daily electricity arrays. k Input the formula for calculating the average daily electricity consumption of the initial typical day. To obtain multiple sets of initial typical daily electricity consumption average data V k ;

[0233] Clustering unit 603 is specifically used for clustering regression models. For any set of target daily electricity data D k-7n and corresponding to D k-7n Target time-of-use electricity data P (k-7n)i Perform clustering to select D k-7n The time-of-use power consumption array;

[0234] Acquisition unit 601, specifically used for obtaining D k-7n Get the daily average hourly electricity consumption data from the time-of-use electricity consumption array.

[0235] Calculation unit 606 is specifically used to calculate the daily time-of-use average electricity consumption data. and V k Substitute the type of day-to-day weighting coefficient calculation formula Calculations are performed to obtain the type-day time-sharing weight coefficient data X. ki ;

[0236] Where n is greater than or equal to 0, k is the date, and D k For the target daily electricity consumption data on day k, D k-7n For the target daily electricity consumption data on day k-7n, P (k-7n)i X represents the target time-of-use electricity data for the i-th time point on day k-7n. ki This refers to the time-sharing weight coefficient data for the type of day at the i-th time point on day k. This represents the average daily hourly electricity consumption data for the i-th time point.

[0237] Optionally,

[0238] Input unit 602 is specifically used to input the target typical daily average electricity consumption data V w+1 and typical daily coefficient data X k Input daily electricity prediction model D k+7 =X k V w+1 To analyze daily electricity consumption data D in short-term load k+7 Make predictions;

[0239] Input unit 602 is specifically used to input daily electricity consumption data D k+7 and type daily time-sharing weight coefficient data X ki Input time-of-use electricity prediction model P (k+7)i =Dk+7 X ki To analyze time-of-use electricity data P in short-term load (k+7)i Make predictions;

[0240] Where k is the date, V w+1 X represents the average daily electricity consumption data for the target typical day in group w+1. k D represents the typical daily coefficient data for day k. k+7 For the target daily electricity consumption data on day k+7, X ki For the type-day time-sharing weight coefficient data of the i-th time point on the k-th day, P (k+7)i This is the hourly electricity consumption data for the i-th time point on day k+7.

[0241] Please refer to the following: Figure 7 The structural schematic diagram of a load forecasting device disclosed in this application includes:

[0242] Central processing unit 701, memory 705, input / output interface 704, wired or wireless network interface 703, and power supply 702;

[0243] Memory 705 is either a short-term storage memory or a persistent storage memory;

[0244] The central processing unit 701 is configured to communicate with the memory 705 and execute instructions stored in the memory 705 to perform the aforementioned operations. Figure 1 or Figure 2 The method in the illustrated embodiment.

[0245] This application also provides a chip system, characterized in that the chip system includes at least one processor and a communication interface, the communication interface and the at least one processor are interconnected via a circuit, and the at least one processor is used to run computer programs or instructions to perform the aforementioned... Figure 1 or Figure 2 The method in the illustrated embodiment.

[0246] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0247] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0248] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0249] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0250] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A load forecasting method, characterized in that, The method includes: Acquire target sample data of the user's historical electricity consumption, the target sample data including multiple target daily electricity consumption data and target time-of-use electricity consumption data corresponding to each target daily electricity consumption data; The target daily electricity consumption data is input into a pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily electricity consumption average data. Multiple sets of the average daily electricity consumption data of the intermediate typical days are input into a pre-trained linear regression model to obtain the average daily electricity consumption data of the target typical days; Clustering regression model is used to cluster the target time-of-use electricity data and the target daily electricity data corresponding to the target time-of-use electricity data to obtain type daily time-of-use weight coefficient data. Short-term load is predicted using the target typical daily average electricity consumption data, the typical daily coefficient data, and the type day time-of-use weighting coefficient data; The typical daily load factor regression model includes an initial typical daily average electricity consumption calculation formula, an intermediate typical daily average electricity consumption calculation formula, and a typical daily coefficient calculation formula; the step of inputting the target daily electricity consumption data into the pre-trained typical daily load factor regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily average electricity consumption data includes: Select , , , , , , As a typical daily electricity consumption array; The different typical daily electricity consumption arrays Input the formula for calculating the initial typical daily average electricity consumption. To obtain multiple sets of initial typical daily electricity consumption average data. ; Multiple groups of the above Input the formula for calculating the average daily electricity consumption of the intermediate typical days. In order to obtain multiple sets of the average daily electricity consumption data of the intermediate typical days. ; Any group described and the aforementioned Input the typical daily coefficient calculation formula The typical daily coefficient data are obtained. ; Among them, the For the date, the stated For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first Typical daily coefficient data; The step of inputting multiple sets of intermediate typical daily electricity consumption average data into a pre-trained linear regression model to obtain target typical daily electricity consumption average data includes: Multiple sets of the average daily electricity consumption data of the intermediate typical values Input the linear regression model The linear regression model is trained to obtain the target typical daily average electricity consumption data. ; Among them, the The The The The The model coefficients of the linear regression model are... For the first The average daily electricity consumption data of the group, the For the first The target typical daily average electricity consumption data for the group; Before clustering the target time-of-use electricity data and the corresponding target daily electricity data using a clustering regression model to obtain the type-day time-of-use weight coefficient data, the method further includes: The target daily electricity data of different typical daily electricity arrays Input the formula for calculating the average daily electricity consumption of the initial typical day. To obtain multiple sets of initial typical daily electricity consumption average data. ; The step of clustering the target time-of-use electricity data and the corresponding target daily electricity data using a clustering regression model to obtain type-day time-of-use weight coefficient data includes: Using cluster regression model For any set of target daily electricity data and corresponding to the The target time-of-use electricity data Perform clustering to select the... The time-of-use power consumption array; Through the above Get the daily average hourly electricity consumption data from the time-of-use electricity consumption array. ; The daily hourly average electricity consumption data of the aforementioned type and the aforementioned Substitute the type of day-to-day weighting coefficient calculation formula Calculations are performed to obtain the time-sharing weighting coefficient data for the aforementioned type. ; Among them, the Greater than or equal to 0, the For the date, the stated For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The day of The target time-of-use electricity data at each point in time, the For the first The day of The type of daily time-sharing weight coefficient data at each time point, the For the first Daily hourly average electricity consumption data for each time point; The method of predicting short-term load using the target typical daily average electricity consumption data, the typical daily coefficient data, and the type day time-of-use weighting coefficient data includes: The target typical daily electricity consumption average data and the typical daily coefficient data Input daily electricity prediction model To analyze daily electricity consumption data in short-term load. Make predictions; The daily electricity data and the daily time-sharing weighting coefficient data of the aforementioned type Input Time-of-Use Electricity Prediction Model To analyze time-of-use electricity data during short-term load periods. Make predictions; Among them, the For the date, the stated For the first The target typical daily average electricity consumption data of the group, the For the first Typical daily coefficient data for the day, the For the first The target daily electricity consumption data, the For the first The day of The type of daily time-sharing weight coefficient data at each time point, the For the first The day of Hourly electricity data at each point in time.

2. The load forecasting method according to claim 1, characterized in that, Before obtaining the target sample data of users' historical electricity consumption, the method further includes: Obtain initial sample data of the user's historical electricity consumption, the initial sample data including multiple initial time-of-use electricity data; Determine whether the initial time-of-use electricity data meets the preset conditions; If so, the initial time-of-use power data is input into the initial power correction model to train the initial power correction model; The target sample data for obtaining users' historical electricity consumption includes: The initial time-of-use power data is input into the trained initial power correction model to obtain multiple sets of the target time-of-use power data. Calculate the target time-of-use electricity data for any set of target time-of-use electricity data to obtain the target daily electricity data corresponding to any set of target time-of-use electricity data.

3. The load forecasting method according to claim 1 or 2, characterized in that, Before obtaining the target sample data of users' historical electricity consumption, the method further includes: Obtain initial sample data of the user's historical electricity consumption, the initial sample data including multiple initial time-of-use electricity consumption data. ; Determine the Is it less than or equal to 0? Is it greater than 0? Is it greater than 0? Or, judge Is it greater than The ; If so, the above Input the initial power correction model In order to train the initial power correction model; The Input the trained initial charge correction model To obtain multiple sets of the target time-of-use electricity data; Calculate the target daily electricity consumption data corresponding to any set of target time-of-use electricity consumption data. ; Among them, the For the date, the stated For the first At each point in time, the For the first The day of The target time-of-use electricity data at each point in time, the For the first The day of The target time-of-use electricity data at each point in time, the For the first The day of The target time-of-use electricity data at each point in time, the For the first The day of The target time-of-use electricity data at each point in time, the For the first The day of The target time-of-use electricity data at each time point, where A and B are model coefficients of the initial electricity correction model, and the... For the first The target daily electricity consumption data.

4. A load forecasting system, characterized in that, The system includes: The acquisition unit is used to acquire target sample data of the user's historical electricity consumption. The target sample data includes multiple target daily electricity consumption data and target time-of-use electricity consumption data corresponding to each target daily electricity consumption data. The input unit is used to input the target daily electricity consumption data into a pre-trained typical daily load coefficient regression model to obtain typical daily coefficient data corresponding to the target daily electricity consumption data and multiple sets of intermediate typical daily electricity consumption average data. The input unit is also used to input multiple sets of intermediate typical daily electricity consumption average data into a pre-trained linear regression model to obtain target typical daily electricity consumption average data. Clustering unit, used to cluster the target time-of-use electricity data and the target daily electricity data corresponding to the target time-of-use electricity data through a clustering regression model, so as to obtain type daily time-of-use weight coefficient data; The prediction unit is used to predict short-term load using the target typical daily average electricity consumption data, the typical daily coefficient data, and the type day time-of-use weighting coefficient data; The system further includes: a selection unit; The selection unit is used to select , , , , , , As a typical daily electricity consumption array; The input unit is used to input the different typical daily electricity consumption arrays. Input the formula for calculating the initial typical daily average electricity consumption. To obtain multiple sets of initial typical daily electricity consumption average data. ; The input unit is specifically used to input multiple sets of the above. Input the formula for calculating the average daily electricity consumption of the intermediate typical days. In order to obtain multiple sets of the average daily electricity consumption data of the intermediate typical days. ; The input unit is specifically used to input any set of the above. and the aforementioned Input the typical daily coefficient calculation formula The typical daily coefficient data are obtained. ; Among them, the For the date, the stated For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first Typical daily coefficient data; The system includes: The input unit is specifically used to process multiple sets of the intermediate typical daily electricity consumption average data. Input the linear regression model The linear regression model is trained to obtain the target typical daily average electricity consumption data. ; Among them, the The The The The The model coefficients of the linear regression model are... For the first The average daily electricity consumption data of the group, the For the first The target typical daily average electricity consumption data for the group; The system also includes: a computing unit; The input unit is specifically used to input the target daily electricity data from different typical daily electricity arrays. Input the formula for calculating the average daily electricity consumption of the initial typical day. To obtain multiple sets of initial typical daily electricity consumption average data. ; The clustering unit is specifically used to perform clustering regression model For any set of target daily electricity data and corresponding to the The target time-of-use electricity data Perform clustering to select the... The time-of-use power consumption array; The acquisition unit is specifically used to obtain the information through the... Get the daily average hourly electricity consumption data from the time-of-use electricity consumption array. ; The calculation unit is specifically used to process the daily hourly average electricity consumption data of the aforementioned type. and the aforementioned Substitute the type of day-to-day weighting coefficient calculation formula Calculations are performed to obtain the time-sharing weighting coefficient data for the aforementioned type. ; Among them, the Greater than or equal to 0, the For the date, the stated For the first The target daily electricity consumption data, the For the first The target daily electricity consumption data, the For the first The day of The target time-of-use electricity data at each point in time, the For the first The day of The type of daily time-sharing weight coefficient data at each time point, the For the first Daily hourly average electricity consumption data for each time point; The input unit is specifically used to input the target typical daily average electricity consumption data. and the typical daily coefficient data Input daily electricity prediction model To analyze daily electricity consumption data in short-term load. Make predictions; The input unit is specifically used to input the daily electricity data. and the daily time-sharing weighting coefficient data of the aforementioned type Input Time-of-Use Electricity Prediction Model To analyze time-of-use electricity data during short-term load periods. Make predictions; Among them, the For the date, the stated For the first The target typical daily average electricity consumption data of the group, the For the first Typical daily coefficient data for the day, the For the first The target daily electricity consumption data, the For the first The day of The type of daily time-sharing weight coefficient data at each time point, the For the first The day of Hourly electricity data at each point in time.

5. A load forecasting device, characterized in that, The device includes: Central processing unit, memory, input / output interfaces, wired or wireless network interfaces, and power supply; The memory is either a short-term storage memory or a persistent storage memory; The central processing unit is configured to communicate with the memory and execute instructions in the memory to perform the method according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1 to 3.

Images