A short-term power load forecasting method and forecasting system

Abstract

The present invention discloses a short-term power load forecasting method and forecasting system, which comprises the following steps: obtaining a first sequence data set; dividing the first sequence data set into a plurality of second sequence data sets; selecting any second sequence data set, processing the second sequence data set by a differential autoregressive moving average method, and obtaining first power load sequence data; processing the second sequence data set by a BP neural network method, and obtaining second power load sequence data; constructing a first objective function, and processing the function by a chaotic particle swarm algorithm, and obtaining a first weight coefficient and a second weight coefficient; traversing all second sequence data sets, and obtaining a plurality of first weight coefficients and a plurality of second weight coefficients; calculating and obtaining the first weight coefficient and the second weight coefficient of the n+1th day; and calculating and obtaining the power load data of the n+1th day. The present invention has the beneficial effects of improving the accuracy of the prediction of the power load sequence data and increasing the precision of the power load sequence data.

Description

Technical Field

[0001] The present invention relates to the technical field of power load forecasting, and in particular to a short-term power load forecasting method and forecasting system. Background Art

[0002] The continuous development of new power systems has increased the difficulty of flexible power system regulation. Short-term power load forecasting plays a crucial role in power system control, grid security, economic operation, system planning, optimal dispatch, and power quality. It is a key component of the dispatching and control system. Therefore, the importance of short-term power load forecasting in power systems has further increased.

[0003] Short-term load forecasting technologies mainly fall into two categories: statistical methods and machine learning methods. Both use a single method to predict future power load data. Since power loads in new power systems are more susceptible to factors such as the access of power electronic devices, temperature and weather changes, and holiday cycles, they have more significant nonlinearity, non-stationarity, and randomness. As a result, it is difficult to accurately reflect these load characteristics using a single forecasting method, resulting in low accuracy in the predicted power load series data.

[0004] In view of this, this application is hereby filed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is that the existing technology adopts a single method to predict power load data, resulting in low accuracy of the predicted power load data. The purpose is to provide a short-term power load forecasting method and forecasting system to improve the accuracy and precision of the power load data forecast.

[0006] The present invention is achieved through the following technical solutions:

[0007] A short-term power load forecasting method, the forecasting method steps comprising:

[0008] Acquire a first sequence data set, where the first sequence data set is historically collected n days of power load sequence data;

[0009] Divide the first sequence data set into a plurality of second sequence data sets, where the second sequence data sets are power load sequence data collected from day 1 to day i, where i<=n;

[0010] Select any second sequence data set, process the second sequence data set using the differential autoregressive moving average method, and obtain the first power load sequence data of the i+1th day;

[0011] The second sequence data set is processed using a BP neural network method to obtain the second power load sequence data of the i+1th day;

[0012] Constructing a first objective function, and processing the first power load sequence data and the second power load sequence data using a chaotic particle swarm optimization algorithm to obtain a first weight coefficient for the i+1th day and a second weight coefficient for the i+1th day, traversing all second sequence data sets to obtain a plurality of first weight coefficients and a plurality of second weight coefficients;

[0013] Calculating a first weight coefficient for day (n+1) and a second weight coefficient for day (n+1) based on the first weight coefficients and the second weight coefficients;

[0014] The power load data of the n+1 day is calculated based on the first weight coefficient of the n+1 day and the second weight coefficient of the n+1 day.

[0015] Traditionally, in the prediction of power load sequence data, statistical methods or machine learning methods are used to predict the power load sequence data separately. However, when the traditional single method is used to predict the power load data, it is usually affected by factors such as the access of power electronic devices, temperature and weather changes, and holiday cycles, which will lead to low accuracy of the final predicted power load sequence data and inaccurate prediction results. The present invention provides a short-term power load forecasting method, which combines the differential autoregressive moving average method with the BP neural network model, calculates the weight coefficient under each mode, and predicts the power load sequence data based on the weight coefficient. The two methods are combined to improve the accuracy of the prediction of the power load sequence data and increase the precision of the power load sequence data.

[0016] Preferably, the specific steps of obtaining the power load data on the (n+1)th day include:

[0017] The first sequence data set is processed using a differential autoregressive moving average method to obtain first power load sequence data of the n+1th day;

[0018] The first sequence data set is processed using a BP neural network method to obtain second power load sequence data for day (n+1);

[0019] The power load data for the (n+1)th day is obtained based on the parameters of the first power load sequence data, the second power load sequence data, the first weight coefficient, and the second weight coefficient.

[0020] Preferably, the calculating of the first weight coefficient for the n+1th day and the second weight coefficient for the n+1th day based on the plurality of the first weight coefficients and the second weight coefficients specifically includes:

[0021] Calculating the proportion of each of the first weight coefficients in the predicted days;

[0022] Based on the occupancy ratio, a first weight coefficient for day (n+1) is calculated;

[0023] The method for calculating the second weight coefficient on the n+1th day is the same as the method for calculating the first weight coefficient on the n+1th day.

[0024] Preferably, the specific expression of the first objective function is:

[0025] f(w A,i+1 ,w B,i+1 )=|w A,i+1 S A,i+1 +w B,i+1 S B,i+1 -S r,i+1 |

[0026] stw A,i+1 +w B,i+1 =1

[0027] w A,i+1 is the weight coefficient of the forecast result of the i+1th day obtained by the difference autoregressive moving average method, w B,i+1 is the weight coefficient of the prediction result of the i+1th day obtained by the BP neural network, S r,i+1 is the actual power load data sequence of day i+1, S A,i+1 is the first power load series data of the forecast result of day i+1 obtained by the differential autoregressive moving average method, S B,i+1 The second power load sequence data of the prediction result for the i+1th day obtained by the BP neural network.

[0028] Preferably, the first sequence data set collects 16 days of historical power load sequence data, where n=16.

[0029] Preferably, the first sequence data is planned to be divided into three second sequence data sets, which are respectively divided into the power load sequence data set collected from the 1st to the 5th day, the power load sequence data set collected from the 1st to the 10th day, and the power load sequence data set collected from the 1st to the 15th day.

[0030] Preferably, the specific expression of the power load data is:

[0031] S p,17 =w A,17 S A,17 +w B,17 S B,17

[0032] S p,17 is the power load series data set for the 17th day of forecast, w A,17is the first weight coefficient for the 17th day of prediction, S A,17 is the 17th day power load series data predicted by the differential autoregressive moving average method, w B,17 is the second weight coefficient for the 17th day of prediction, S B,17 This is the power load series data for the 17th day predicted using the BP neural network.

[0033] Preferably, the w A,17 The specific expression is:

[0034]

[0035] w A,6 is the first weight coefficient of the 6th day calculated by the difference autoregressive moving average method, w A,11 is the first weight coefficient of the 11th day predicted by the difference autoregressive moving average method, w A,16 is the first weight coefficient of the 16th day predicted by the difference autoregressive moving average method.

[0036] Preferably, the w B,17 The specific expression is:

[0037]

[0038] w B,6 is the first weight coefficient calculated on the 6th day using the BP neural network method, w B,11 is the first weight coefficient of the 11th day predicted by the BP neural network method, w B,16 is the first weight coefficient of the 16th day predicted by the BP neural network method.

[0039] The present invention also provides a short-term power load forecasting system, which includes a data acquisition module, a data set partitioning module, a first processing module, a second processing module, a weight coefficient calculation module, a first calculation module, and a second calculation module;

[0040] The data acquisition module is used to acquire a first sequence data set, where the first sequence data set is historically collected n days of power load sequence data;

[0041] The data set division module is configured to divide the first sequence data set into a plurality of second sequence data sets, where the second sequence data sets are power load sequence data collected from day 1 to day i, where i<=n;

[0042] The first processing module is used to select any second sequence data set and process the second sequence data set using a differential autoregressive moving average method to obtain first power load sequence data of the i+1th day;

[0043] The second processing module is used to process the second sequence data set using a BP neural network method to obtain the second power load sequence data of the i+1th day;

[0044] The weight coefficient calculation module is used to construct a first objective function and process the first power load sequence data and the second power load sequence data using a chaotic particle swarm algorithm to obtain a first weight coefficient for the i+1th day and a second weight coefficient for the i+1th day, and traverse all second sequence data sets to obtain a plurality of first weight coefficients and a plurality of second weight coefficients;

[0045] The first calculation module is configured to calculate a first weight coefficient for the n+1th day and a second weight coefficient for the n+1th day based on the first weight coefficients and the second weight coefficients;

[0046] The second calculation module is used to calculate and obtain the power load data of the n+1th day based on the first weight coefficient of the n+1th day and the second weight coefficient of the n+1th day.

[0047] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0048] An embodiment of the present invention provides a short-term power load forecasting method and forecasting system, which combines the differential autoregressive moving average method with the BP neural network model, calculates the weight coefficient under each mode, and predicts the power load sequence data based on the weight coefficient. The combination of the two methods improves the accuracy of the prediction of the power load sequence data and increases the precision of the power load sequence data. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without making any creative efforts.

[0050] Figure 1 Schematic diagram of the prediction method

[0051] Figure 2 Schematic diagram of the prediction system

[0052] Figure 3 is the actual power load data series

[0053] Figure 4 Comparison chart of predicted values ​​and actual values DETAILED DESCRIPTION

[0054] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0055] In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one skilled in the art that these specific details are not necessarily required to practice the present invention. In other embodiments, well-known structures, circuits, materials, or methods are not described in detail to avoid obscuring the present invention.

[0056] Throughout this specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Therefore, appearances of the phrases "one embodiment," "an embodiment," "an example," or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combinations and / or subcombinations. Furthermore, it will be understood by those of ordinary skill in the art that the figures provided herein are for illustrative purposes only and are not necessarily drawn to scale. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0057] In the description of the present invention, the terms "front", "back", "left", "right", "up", "down", "vertical", "horizontal", "high", "low", "inside", "outside", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the scope of protection of the present invention.

[0058] Example 1

[0059] This embodiment 1 discloses a short-term power load forecasting method. In this embodiment, the method is to obtain 16 days of historical power load data to estimate the power load data of the 17th day. Figure 1 As shown, the prediction method steps include:

[0060] S1: Acquire a first sequence data set, which is historically collected n-day power load sequence data;

[0061] The first sequence data set collects 16 days of historical power load sequence data, n=16.

[0062] The power load data from September 1 to September 16, 2018, is collected as training data to predict the power load data sequence S on September 17. p,17 The actual power load data from the 1st day to the 16th day are S r,1 、S r,2 、S r,3 、S r,4 、S r,5 、S r,6 、S r,7 、S r,8 、S r,9 、S r,10 、S r,11 、S r,12 、S r,13 、S r,14 、S r,15 、S r,16 , the data sequence is as follows Figure 3 As shown;

[0063] S2: Divide the first sequence data set into a plurality of second sequence data sets, where the second sequence data set is the power load sequence data collected from day 1 to day i, where i<=n;

[0064] The first sequence data is planned to be divided into three second sequence data sets, which are respectively divided into the power load sequence data set collected from the 1st to the 5th day, the power load sequence data set collected from the 1st to the 10th day, and the power load sequence data set collected from the 1st to the 15th day.

[0065] S3: Select any second sequence data set, and process the second sequence data set using the differential autoregressive moving average method to obtain the first power load sequence data of the i+1th day;

[0066] Based on the power load data series from the 1st to the 5th day, the power load data series S on the 6th day is predicted using the Autoregressive Integrated Moving Average (ARIMA) model. A,6 Based on the power load data series from the 1st to the 10th day, use ARIMA to predict the power load data series S on the 11th day. A,11 Based on the power load data series from the 1st to the 15th day, ARIMA is used to predict the power load data series S on the 16th day. A,16 ;

[0067] S4: Processing the second sequence data set using a BP neural network method to obtain the second power load sequence data for day i+1;

[0068] Based on the power load data sequence from the 1st to the 5th day, the BP neural network algorithm is used to predict the power load data sequence S of the 6th day. B,6Based on the power load data sequence from the 1st to the 10th day, the BP neural network algorithm is used to predict the power load data sequence S on the 11th day. B,11 Based on the power load data sequence from the 1st to the 15th day, the BP neural network algorithm is used to predict the power load data sequence S on the 16th day. B,16 ;

[0069] S5: Constructing a first objective function and processing the first power load sequence data and the second power load sequence data using a chaotic particle swarm optimization algorithm to obtain a first weight coefficient for the i+1th day and a second weight coefficient for the i+1th day, traversing all second sequence data sets to obtain a plurality of first weight coefficients and a plurality of second weight coefficients;

[0070] The specific expression of the first objective function is:

[0071] f(w A,i+1 ,w B,i+1 )=|w A,i+1 S A,i+1 +w B,i+1 S B,i+1 -S r,i+1 |

[0072] stw A,i+1 +w B,i+1 =1

[0073] w A,i+1 is the weight coefficient of the forecast result of the i+1th day obtained by the difference autoregressive moving average method, w B,i+1 is the weight coefficient of the prediction result of the i+1th day obtained by the BP neural network, S r,i+1 is the actual power load data sequence of day i+1, S A,i+1 is the first power load series data of the forecast result of day i+1 obtained by the differential autoregressive moving average method, S B,i+1 The second power load sequence data of the prediction result for the i+1th day obtained by the BP neural network.

[0074] S6: Calculate a first weight coefficient for the (n+1)th day and a second weight coefficient for the (n+1)th day based on the first weight coefficients and the second weight coefficients;

[0075] The calculating of the first weight coefficient for the n+1th day and the second weight coefficient for the n+1th day based on the plurality of the first weight coefficients and the second weight coefficients specifically includes:

[0076] Calculating the proportion of each of the first weight coefficients in the predicted days;

[0077] Based on the occupancy ratio, a first weight coefficient for day (n+1) is calculated;

[0078] The method for calculating the second weight coefficient on the n+1th day is the same as the method for calculating the first weight coefficient on the n+1th day.

[0079] S7: Calculate and obtain the power load data of the n+1th day based on the first weight coefficient of the n+1th day and the second weight coefficient of the n+1th day.

[0080] The specific steps of obtaining the power load data on day (n+1) include:

[0081] The first sequence data set is processed using a differential autoregressive moving average method to obtain first power load sequence data of the n+1th day;

[0082] The first sequence data set is processed using a BP neural network method to obtain second power load sequence data for day (n+1);

[0083] The power load data for the (n+1)th day is obtained based on the parameters of the first power load sequence data, the second power load sequence data, the first weight coefficient, and the second weight coefficient.

[0084] Based on the power load data series from the 1st to the 16th day, ARIMA is used to predict the power load data series S on the 17th day. A,17 Based on the power load data sequence from the 1st to the 16th day, the BP neural network algorithm is used to predict the power load data sequence S on the 17th day. B,17 ;

[0085] The specific expression of the power load data is:

[0086] S p,17 =w A,17 S A,17 +w B,17 S B,17

[0087] S p,17 is the power load series data set for the 17th day of forecast, w A,17 is the first weight coefficient for the 17th day of prediction, S A,17 is the 17th day power load series data predicted by the differential autoregressive moving average method, w B,17 is the second weight coefficient for the 17th day of prediction, S B,17 This is the power load series data for the 17th day predicted using the BP neural network.

[0088] The w A,17 The specific expression is:

[0089]

[0090] w A,6 is the first weight coefficient of the 6th day calculated by the difference autoregressive moving average method, w A,11 is the first weight coefficient of the 11th day predicted by the difference autoregressive moving average method, w A,16 is the first weight coefficient of the 16th day predicted by the difference autoregressive moving average method.

[0091] The w B,17 The specific expression is:

[0092]

[0093] w B,6 is the first weight coefficient calculated on the 6th day using the BP neural network method, w B,11 is the first weight coefficient of the 11th day predicted by the BP neural network method, w B,16 is the first weight coefficient of the 16th day predicted by the BP neural network method.

[0094] The comparison between the predicted value and the actual value of the short-term power load forecasting method based on the combined model of the present invention is as follows: Figure 4 As shown in the figure, it can be seen that the short-term power load forecasting method based on the combined model can better predict the short-term power load.

[0095] This embodiment discloses a short-term power load forecasting method, which combines the differential autoregressive moving average method with the BP neural network model, calculates the weight coefficient under each mode, and predicts the power load sequence data based on the weight coefficient. The combination of the two methods improves the accuracy of the prediction of the power load sequence data and increases the precision of the power load sequence data.

[0096] Example 2

[0097] This embodiment discloses a short-term power load forecasting system. This embodiment is to implement the forecasting method in the first embodiment. Figure 2 As shown, the prediction system includes a data acquisition module, a data set partitioning module, a first processing module, a second processing module, a weight coefficient calculation module, a first calculation module and a second calculation module;

[0098] The data acquisition module is used to acquire a first sequence data set, where the first sequence data set is historically collected n days of power load sequence data;

[0099] The data set division module is configured to divide the first sequence data set into a plurality of second sequence data sets, where the second sequence data sets are power load sequence data collected from day 1 to day i, where i<=n;

[0100] The first processing module is used to select any second sequence data set and process the second sequence data set using a differential autoregressive moving average method to obtain first power load sequence data of the i+1th day;

[0101] The second processing module is used to process the second sequence data set using a BP neural network method to obtain the second power load sequence data of the i+1th day;

[0102] The weight coefficient calculation module is used to construct a first objective function and process the first power load sequence data and the second power load sequence data using a chaotic particle swarm algorithm to obtain a first weight coefficient for the i+1th day and a second weight coefficient for the i+1th day, and traverse all second sequence data sets to obtain a plurality of first weight coefficients and a plurality of second weight coefficients;

[0103] The first calculation module is configured to calculate a first weight coefficient for the n+1th day and a second weight coefficient for the n+1th day based on the first weight coefficients and the second weight coefficients;

[0104] The second calculation module is used to calculate and obtain the power load data of the n+1th day based on the first weight coefficient of the n+1th day and the second weight coefficient of the n+1th day.

[0105] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A short-term power load forecasting method, characterized in that: The prediction method steps include: Acquire a first sequence data set, where the first sequence data set is historically collected n days of power load sequence data; Divide the first sequence data set into a plurality of second sequence data sets, where the second sequence data sets are power load sequence data collected from day 1 to day i, where i<=n; Select any second sequence data set, process the second sequence data set using the differential autoregressive moving average method, and obtain the first power load sequence data of the i+1th day; The second sequence data set is processed using a BP neural network method to obtain the second power load sequence data of the i+1th day; Constructing a first objective function, and processing the first power load sequence data and the second power load sequence data using a chaotic particle swarm optimization algorithm to obtain a first weight coefficient for the i+1th day and a second weight coefficient for the i+1th day, traversing all second sequence data sets to obtain a plurality of first weight coefficients and a plurality of second weight coefficients; Calculating a first weight coefficient for day (n+1) and a second weight coefficient for day (n+1) based on the first weight coefficients and the second weight coefficients; Calculate and obtain the power load data for day n+1 based on the first weight coefficient for day n+1 and the second weight coefficient for day n+1; The step of calculating the first weight coefficient for the n+1th day and the second weight coefficient for the n+1th day based on the first weight coefficients and the second weight coefficients specifically includes: calculating the proportion of each of the first weight coefficients in the predicted days; and calculating the first weight coefficient for the n+1th day based on the proportions; the method for calculating the second weight coefficient for the n+1th day is the same as the method for calculating the first weight coefficient for the n+1th day. The specific expression of the first objective function is: f(w A,i+1 ,w B,i+1 )=|w A,i+1 S A,i+1 +w B,i+1 S B,i+1 -S r,i+1 | s.t.w A,i+1 +w B,i+1 =1 w A,i+1 is the weight coefficient of the forecast result of the i+1th day obtained by the difference autoregressive moving average method, w B,i+1 is the weight coefficient of the prediction result of the i+1th day obtained by the BP neural network, S r,i+1 is the actual power load data sequence of day i+1, S A,i+1 is the first power load series data of the forecast result of day i+1 obtained by the differential autoregressive moving average method, S B,i+1 The second power load sequence data of the prediction result for the i+1th day obtained by the BP neural network.

2. A short-term power load forecasting method according to claim 1, characterized in that: The specific steps of obtaining the power load data on day (n+1) include: The first sequence data set is processed using a differential autoregressive moving average method to obtain first power load sequence data of the n+1th day; The first sequence data set is processed using a BP neural network method to obtain second power load sequence data for day (n+1); The power load data for the (n+1)th day is obtained based on the parameters of the first power load sequence data, the second power load sequence data, the first weight coefficient, and the second weight coefficient.

3. A short-term power load forecasting method according to claim 1, characterized in that: The first sequence data set collects 16 days of historical power load sequence data, n=16.

4. A short-term power load forecasting method according to claim 3, characterized in that: The first sequence data is planned to be divided into three second sequence data sets, which are respectively divided into the power load sequence data set collected from the 1st to the 5th day, the power load sequence data set collected from the 1st to the 10th day, and the power load sequence data set collected from the 1st to the 15th day.

5. A short-term power load forecasting method according to claim 4, characterized in that: The specific expression of the power load data is: S p,17 =w A,17 S A,17 +w B,17 S B,17 S p,17 is the power load series data set for the 17th day of forecast, w A,17 is the first weight coefficient for the 17th day of prediction, S A,17 is the 17th day power load series data predicted by the differential autoregressive moving average method, w B,17 is the second weight coefficient for the 17th day of prediction, S B,17 This is the power load series data for the 17th day predicted using the BP neural network.

6. A short-term power load forecasting method according to claim 5, characterized in that: The w A,17 The specific expression is: w A,6 is the first weight coefficient of the 6th day calculated by the difference autoregressive moving average method, w A,11 is the first weight coefficient of the 11th day predicted by the difference autoregressive moving average method, w A,16 is the first weight coefficient of the 16th day predicted by the difference autoregressive moving average method.

7. A short-term power load forecasting method according to claim 5, characterized in that: The w B,17 The specific expression is: w B,6 is the first weight coefficient calculated on the 6th day using the BP neural network method, w B,11 is the first weight coefficient of the 11th day predicted by the BP neural network method, w B,16 is the first weight coefficient of the 16th day predicted by the BP neural network method.

8. A short-term power load forecasting system, characterized in that: The prediction system includes a data acquisition module, a data set partitioning module, a first processing module, a second processing module, a weight coefficient calculation module, a first calculation module, and a second calculation module; The data acquisition module is used to acquire a first sequence data set, where the first sequence data set is historically collected n days of power load sequence data; The data set division module is configured to divide the first sequence data set into a plurality of second sequence data sets, where the second sequence data sets are power load sequence data collected from day 1 to day i, where i<=n; The first processing module is used to select any second sequence data set and process the second sequence data set using a differential autoregressive moving average method to obtain first power load sequence data of the i+1th day; The second processing module is used to process the second sequence data set using a BP neural network method to obtain the second power load sequence data of the i+1th day; The weight coefficient calculation module is used to construct a first objective function and process the first power load sequence data and the second power load sequence data using a chaotic particle swarm algorithm to obtain a first weight coefficient for the i+1th day and a second weight coefficient for the i+1th day, and traverse all second sequence data sets to obtain a plurality of first weight coefficients and a plurality of second weight coefficients; The first calculation module is configured to calculate a first weight coefficient for the n+1th day and a second weight coefficient for the n+1th day based on the first weight coefficients and the second weight coefficients; The second calculation module is configured to calculate and obtain the power load data of the n+1 day based on the first weight coefficient of the n+1 day and the second weight coefficient of the n+1 day; The first calculation module is further configured to: calculate the proportion of each first weight coefficient in the predicted days; and calculate the first weight coefficient for the n+1th day based on the proportion; The method for calculating the second weight coefficient on the n+1th day is the same as the method for calculating the first weight coefficient on the n+1th day; The specific expression of the first objective function in the weight coefficient calculation module is: f(w A,i+1 ,w B,i+1 )=|w A,i+1 S A,i+1 +w B,i+1 S B,i+1 -S r,i+1 | s.t.w A,i+1 +w B,i+1 =1 w A,i+1 is the weight coefficient of the forecast result of the i+1th day obtained by the difference autoregressive moving average method, w B,i+1 is the weight coefficient of the prediction result of the i+1th day obtained by the BP neural network, S r,i+1 is the actual power load data sequence of day i+1, S A,i+1 is the first power load series data of the forecast result of day i+1 obtained by the differential autoregressive moving average method, S B,i+1 The second power load sequence data of the prediction result for the i+1th day obtained by the BP neural network.

Images